In-Memory BI Market

QlikTech Confidential


MARKET EVOLUTION

The central problem that organizations face is how to optimize business processes. The past 20 years has seen organizations make significant investments in deploying mechanisms for automating business processes which previously were manual. This investment has single-handedly built Silicon Valley. Organizations have invested trillions in deploying systems to capture, store, and manipulate data in databases and applications instead of paper and filing cabinets. This effort has driven tremendous value in the Information Technology business, giving rise to thousands of software companies and behemoths like SAP, Oracle, IBM, and Microsoft.

As soon as organizations have automated a business process, the next logical question is how to take advantage of this captured data and optimize the process. Process measurement and change in paper based systems is time consuming and slow. The vision of the IT industry is that this organizational change is faster, easier, and less expensive once a process has been automated.

This is the central challenge of the Business Intelligence market – how to make sense of the voluminous data captured in automated systems for the purpose of optimizing business.

In this paper we will discuss how the technology to answer this pressing business challenge has evolved over the past 20 years; the changes which have allowed a new set of solutions to emerge and disrupt the status quo; and the reactions of the traditional competitors to this disruptive change.

The OLAP Tradition

Twenty years ago memory was expensive and processors were slow. The difference in price/performance for both factors is well over a factor of 1000 higher today than it was then. Faced with these constraints developers at the time devised an architectural approach for delivering results of multi-dimensional analysis which relied on pre-calculating fixed analyses. Simply put, they pre-calculated all measures across every possible combination of dimensions. For example, for total sales by sales person and region, the system would calculate total sales for each sales person for each region, and for every union of sales person and region. The results of these calculations were stored and retrieved when an end user requested a particular “analysis.” This is what is traditionally referred to as “calculating the cube” and the “cube” is the mechanism which organizes and stores the



results. Because the results were pre-calculated, regardless of how long in took to calculate the results, the response time from the perspective of the end user was instantaneous.

Today, we have available to us a fundamentally different technology platform on which to build business intelligence. Specifically three things have happened.

First, Moore’s Law has relentlessly beat its drum resulting in processors which are significantly faster today than they were 20 years ago and memory which is significantly less expensive.

Second, is the mainstream availability of 64-bit processors which raises the amount of memory a computer can utilize. A 32-bit processor can use 4 gigabytes of memory at a maximum, and a portion of that must be devoted to the operating system. A 64-bit processor can use 17,179,869,184 gigabytes or 16 exabytes of RAM – a factor of 4 billion more. Of course, the practical limitation of computers available today is much lower, but machines with 40, 80, or even 120 gigabytes of memory are readily available for less than $30,000.

Third, is the shift away from computers with fast processors, to computers with multiple low-power lower-speed processors. The challenge today is keeping computers operating at a reasonable temperature. Intel and AMD’s stated strategy for achieving this goal is to equip computers with many lower power processors working in parallel. Today it is common to find computers with 2, 4, 16, 32 or even 128 processors. In addition, newer processors have multiple “cores” bundled on a single chip.

The Emergence of a New Space

With the aforementioned changes in the underlying hardware/processor platform the business intelligence market finds itself on the cusp of a technology transition. In short, the market is moving from OLAP solutions to increasingly adopt In-Memory business intelligence solutions. To understand In-Memory analysis it is useful to characterize the technology it replaces: MOLAP and ROLAP.

MOLAP is “multidimensional” OLAP. The distinctive feature of MOLAP is that it stores the results of a cube in a multidimensional store. The form and exact nature of this multidimensional store is proprietary and specific to the particular vendor whose tool is being used. Because the storage of the cube is proprietary, MOLAP can provide some unique advantages such as the ability to support writing results back in to a cube (for budgeting) or compression of cubes on disk.

Example MOLAP Vendors: Hyperion, Applix TM1, Palo

Advantages of MOLAP • Fast query performance due to optimized storage, multidimensional indexing and

caching. • It is very compact for low dimension data sets.

Disadvantages of MOLAP • The processing step (data load) can be quite lengthy, especially on large data

volumes.



• MOLAP tools traditionally have difficulty querying models with dimensions with very high cardinality (i.e., millions of unique values).

ROLAP is “relational” OLAP and differs from MOLAP in that the resulting cube is stored in a relational database. Data in a ROLAP implementation is stored in a star (or snowflake) schema which optimizes the layout of the data to be read by a database engine.

Example ROLAP Vendors: Cognos, Microstrategy, Business Objects

Advantages of ROLAP • ROLAP is considered to be more scalable than MOLAP in handling large data

volumes. • Load times are generally much shorter than with the automated MOLAP loads.

Disadvantages of ROLAP • The process of loading data in to star/snowflake schemas is difficult and involves

management of complex ETL code. • ROLAP data stores tend to “explode” in size as many dimensions are added.

In-Memory Analysis (IMOLAP?) differs from ROLAP and MOLAP in that the primary storage mechanism for data to be analyzed is memory. Typically vendors in this space don’t pre-calculated measures, but rather rely on the speed of memory to allow values to be calculated as they are needed. There is significantly less uniformity of approach by vendors in this space – some offer only fast queries and no calculation or UI, others are simply implementations of cubes which are held in memory.

Example In-Memory Vendors: Applix, Panoratio, QlikTech, and Spotfire

Advantages of In-Memory • Memory is significantly faster than disk which results in fast queries and

calculations. • Eliminating building cubes speeds deployment and allows revision to analysis

more quickly • Fast access to queries and aggregates allows new ways to visualize and

manipulate data (such as QlikView’s Association Technology).

Disadvantages of In-Memory • Typically refreshes are time consuming because all data needs to be loaded in to

memory (unless product supports incremental reload). • Without 64-bit technology there is a significant limit to the amount of data that

can be held in memory. • Data is analyzed in memory, not in the underlying store. Therefore data in

memory is always “out of date.” This eliminates the possibility of “real-time” analysis.



Two additional points of detail are useful in characterizing the In-Memory solution: first is additional clarification of the notion of “real-time”; second is how ETL integrates with In-Memory solutions.

Due to their architecture, OLAP solutions are not well suited to handling real time analysis. Real time analysis is defined as: monitoring database traffic as it comes into a database and taking action when certain criteria are met. For instance, watch all orders as they are placed, and send an alert if a big order is placed for a product that is out of stock. Because In-Memory solutions extract data from the underlying source and hold it in memory, by definition the data is out of date (a old as the last load time). ROLAP and MOLAP vendors do not handle this any better, as they also extract data from the underlying system. However, traditional OLAP solutions also can run queries on operational stores as users drill to detail records. In addition, the space of BAM (Business Activity Monitoring) has emerged to handle these requirements directly. QlikView has done two things in this area. First, for customers who need this we've offered an integrated approach where we blend our QlikView with a "real time feed" and put the two UI's together. Second, in 8.0 we will have the ability to enter data into QlikView (i.e. input placeholder fields) and this, combined with macros, could easily allow real time data to be shown inside QlikView.

ETL is really three tasks: extract from the underlying source, transform into a new schema, and load into a new system. Most of the challenge with ETL is transforming data so that it can efficiently be loaded into data warehouses or into cubes. Neither is a useful task for In-Memory analysis, because data need not be stored in either the warehouse or the cube. That said, for QlikView the load script is robust enough to handle most extraction, and many (most) transformations -- the load step is not something that is necessary in In-Memory analysis.

So, with the 64-bit processors lifting the memory limitation on servers, and the shift to low power multi-processor systems a new market for memory powered business intelligence is emerging.

Enter QlikView

QlikView was built with a simple architectural premise – all data should be held in memory, and all calculations should be performed when requested and not prior. Twenty years ago this would have been impossible. In 1993, when QlikTech was founded, it was still a pretty crazy idea. And this approach, for much of QlikTech’s early history, significantly limited the amount data and the scope of analysis possible. But, the trends in the underlying platform have lifted the constraints and now this approach is winning share.

As a solution, QlikView offers three components in an integrated solution:

• Fast Query Engine: loading the data into memory allows QlikView to query, or sub-set, the data instantly to only reveal the data which is relevant to a given user. In addition, QlikView shows users the data which is excluded by a selection.



• On Demand Calculation Engine: charts, graphs, and tables of all types in QlikView are multidimensional analysis. That is, they show one or more measures (metrics, KPIs, expressions, etc.) across one or more dimensions (example: total sales by region). The major difference is that these calculations are performed as the user clicks and never prior.

• Visually Interactive User Interface (UI): QlikView offers hundreds of possible chart and table types and varieties; there are list boxes for navigating dimensions; statistic boxes; and many other UI elements. Every UI element can be clicked on to query. The QlikView UI is the representation of our associative model for data interaction. It says that data that is included in a query should be show, as well as data that is excluded from a query.

Taken together the underlying technology foundation for QlikView is called In-Memory Association Technology. It is referred to as “In-Memory” for the simple reason that it holds all data in memory, and operates (queries and aggregates) that data in memory. The concept of Association refers to the associative mapping between data elements that QlikView performs. This associative model allows users to navigate and use QlikView in the same way their brain thinks about a problem. Humans think about challenges non-linearly, and they think about possibilities included and excluded at the same time. Much in the same way QlikView’s association technology shows information that is included and excluded at the same time, and allows the user a unique path through the data. QlikTech has patents on this technology foundation.

The QlikView solution, because of its unique integrated components and because it operates entirely in memory, offers some unique advantages over traditional OLAP.

• Fast Time-to-Value: with traditional OLAP constructing cubes is time consuming and requires expert skills. This process can take months, and sometimes over a year. In addition the cube must be constructed before it can be calculated, a process which itself can take hours. And, all this must occur before analysis or reporting can be performed – before the user even sees answers to his questions. Because the data is loaded in memory, creating analysis in QlikView takes seconds. There is no pre-definition of what is a dimension, any data is available as a dimension and any data is available as a measure. The time implementing QlikView is spent locating data, and deciding what analysis is interesting or relevant to solving the business question. Typically this process takes a week or two.

• Easy to Use: The entire end user experience in QlikView is driven by the “click.” End users enjoy using QlikView because it works the way their mind does. Each time they want to review the data sliced a new way, they simply click on the data they want to evaluate. Because QlikView operates in memory, with each click all data and measures are recalculated to reflect the selection.



• Powerful: Because queries and calculations are performed in memory they are extremely quick. In addition, QlikView is not constrained by the speed of the underlying source. Even if the underlying data is stored in a system which has poor query performance (for instance a text file), because the data is loaded in memory the performance is always optimal. QlikView also compresses data as it is stored in memory allowing large amounts of data to be stored. Typically we see ten times reduction in size of the data once it’s in memory.

• Flexible: One of the major issues with traditional OLAP is that modifying an analysis requires changing the cube, a process which can take a very long time. In addition this process is typically controlled by IT. With QlikView, viewing analysis by a new dimension or changing a measure can be performed by business professionals in seconds.

The Rise of the Stack Player

In addition to the In-Memory analysis innovation, a second major trend in the BI market is the rise of the “stack player” as a viable business intelligence tool.

The pervading wisdom in the BI market was that in order to analyze data in an operational system, that data must be moved from a transactional database to a new repository and transformed into a different schema. The vendors who provided this capability were sold as incremental tools beyond the underlying database from which they extracted their data. The stack players have not let this lost revenue go unnoticed.

Stack players are vendors who provide infrastructure and tools for the deployment of systems for automating business processes. This can include some or all of the following: hardware, database, application server, business applications, and analytics.

Stand-alone BI vendors are vendors who only provide analytics from the list above. Here the term analytics is used to describe the broadest possible offering used to analyze data in operational systems.

The Great BI Squeeze

This leaves stand-alone BI vendors now face a dual threat. From the bottom they face innovative new technology and from the top they face the bundling of traditional OLAP technology into the offerings of the stack players. The core offering of BI, OLAP technology, is becoming commoditized by being bundled into the stack player’s offering. And, worst yet, the stand-alone BI vendors have missed the boat on building the next generation of BI technology – In-Memory Analysis and Reporting.

At least one half of this squeeze seems unavoidable: the stand-alone BI vendors cannot react to innovation from below because it disrupts their business model. For each of the top three stand-alone BI vendors, well over 50% of their revenue comes from services:

• Cognos: 36.8% of revenue is license software (Q1 2006) • Business Objects: 45% of revenue is license software (Q1 2006) • Microstrategy: 30.4% of revenue is license software (Q1 2006) • Hyperion: 34.6% of revenue is license software (Q1 2006)



Most of that services time and money is spent in multi month (year) engagements with customers creating, deploying, and managing cubes. Adopting an in-memory approach to BI would obviate the need to build and maintain cubes, and thus eliminate a significant portion of 50% of their revenue.

So, stand-alone BI vendors are caught between the commoditization of their technology and the inability to react to a disruptive technology which threatens their business model.

The Reaction of Stand-Alone BI

As a group these vendors have done very little to combat the rise of In-Memory BI, and have reacted predictably to the stack players.

Cognos/Microstrategy/Business Objects: None of these vendors has made any investments in In-Memory technology. They are strongly committed to their path of traditional OLAP.

In reaction to the stack-players each of them has made significant investments in broadening their solutions set. Now they provide the full set of functionality from OLAP ad-hoc analysis to static reporting. They have also invested significant time building ”analytic applications”. These are templates of analysis which solve a specific analytical problem and, in some cases, the underlying data model to support them. This makes them more complete as a solution than the stack players. Their bet seems to be that, but focusing entirely on BI, they can build a richer more complete solution than the stack-players – and that “good enough” isn’t.

Hyperion: Hyperion’s recent moves suggest that it recognizes that In-Memory (or, more generally, some set of new technology) is necessary to keep innovating. They have recently hired Howard Dresner and Frank Buytendijk from Gartner, both well versed in In-Memory technology (especially as expressed by QlikTech) and new technology as well.

Hyperion’s reaction to the stack-players has been the same as the above vendors.

The Reaction of the Stack Players

Generally the stack-players have been much more receptive to looking at new technology in the BI space, presumably to differentiate their offerings from the stand-alone vendors.

IBM: IBM has done very little in the In-Memory space, and generally is the stack-vendor with the least capable business intelligence software offering. They recently made a massive investment to purchase Ascential Software which provides technology to load data from operational systems into other systems or into cubes. This offering fits more cleanly with IBM’s moves in the data integration space than with indicating a focus on BI. That said, IBM has shown interest in In-Memory technology as evidenced by recent blog entries by their employees.

SAP: SAP has made significant investments in In-Memory technology and has a product called the SAP BI Accelerator which is based on in memory full text search technology.



Their product, however, only applies the speed and power of in memory to cubes built in SAP, therefore significantly limiting its usefulness. SAP’s marketing message behind this technology, however, mirrors very closely (almost too closely) QlikTech’s messaging around QlikView. Too bad the product doesn’t meet the marketing message.

Oracle: Oracle has significant investments in In-Memory technology, but only because they acquired Siebel Analytics. Oracle’s base technology is very standard OLAP. Siebel Analytics is an In-Memory reporting tool which uses memory to combine the results of queries to underlying data sources. The significant downside to this approach is that the performance of the system is directly related to the performance of the underlying queries, which is often terrible. In short, they don’t take advantage of the speed of memory.

HP: HP has little to offer in BI in general, and certainly nothing based on In-Memory technology. They are discussed again below in the hardware platform section.

Microsoft: Microsoft’s capabilities in In-Memory technology really focus around in memory caching in SQL Server, their database platform. Their analytics solution, Microsoft Analysis Services, is squarely OLAP, and very standard at that. That said, they have address the visualization layer of their analysis platform with the acquisition of ProClarity. ProClarity is not an in memory technology, but is a fantastic/pretty front end to Microsoft Analysis Services.

The Reaction of the Hardware Platforms

The hardware platforms are, understandably, very interested in In-Memory technology. Intel, AMD, HP, and IBM (to a lesser extent Dell) have moved their entire technology stack to 64-bit technology, but have not found a convincing reason for organizations to adopt the technology. The best examples they have found are CAD/CAM, image editing, and video – none of which are enterprise application, and none of which affect more than a very small group of power users. In-Memory Analysis has the opportunity to change that because it is an application which reaches the entire enterprise, and because it allows functionality which is physically impossible in a 32-bit world.

One challenge in working with HP in particular is their obsession with Itanium. At this point it is quite clear that the Itanium platform is the second child to Xeon. It is far too expensive, and narrowly adopted. However, when working with HP it is necessary to ensure that everything is balanced between Itanium and Xeon. That said, for QlikView in particular, when data volumes get large or user counts get large there is no other platform which can scale up as well as Itanium. We expect that with time Xeon will reach Itanium in terms of scalability, but not in the short term.

The In-Memory Market Players

Any new market must have competitors, and typically a clear market leader. The In-Memory Market is defined as having four technology components (detailed above): leverage of 64-bit hardware; performing queries in memory; aggregating data in real-time; and utilizing a strong end-user visualization UI. In addition players in this market can gain traction in customer deployments and market awareness.



QlikTech has emerged as the clear market leader. QlikView has been built from the ground up to take advantage of 64-bit/multicore technology. For each processor platform (32-bit, Xeon, Itanium) processor-specific assembly code is generated to achieve unprecedented performance. Rather than simply recompiling for the 64-bit environment, QlikView is natively designed for the new hardware platforms. QlikView holds all data in memory, and queries that memory store resulting in very fast query times. All aggregation is done directly in memory, and aggregates are never stored to disk. QlikView’s user interface is visually interactive, and uses a simple but powerful green-white-grey metaphor that anyone can understand.

As of September 2006 QlikTech has 4,440 customers in 63 countries and has added 2,495 net new customers in the prior 12 months. This is both more In-Memory Analysis customers than anyone else, but also a significantly higher growth rate. QlikTech is also recognized as the fastest growing BI software company by IDC, and in press reports as the leader in In-Memory Analysis.

SAP BI Accelerator: SAP’s BI Accelerator is deployed exclusively on 64-bit XEON blade servers from IBM or HP. It runs on a proprietary version of Linux. Thus, we can assume that they are well tuned for the 64-bit environment. The underlying technology for BI Accelerator is the SAP TREX full text search engine. This suggests that much of the querying is either done entirely in memory or close to it. The product executes queries exclusively on top of SAP BW cubes and, when an aggregate cannot be found in the cube, builds that aggregate on the fly. There is no user interface other than SAP’s standard BI tools which are weak on visualization.

On the traction side, SAP has made significant noise in the market, but their product functionality does not match their marketing pitch. They currently have around 30 customers which, given prior SAP history, suggests around 10-15 of them are “real”.

Oracle (Siebel) Analytics: Oracle acquired Siebel Analytics which Siebel acquired from a company called nQuire. Siebel Analytics executes queries directly on the underlying



source, so it is entirely reliant on the performance of the underlying system. However, once the query executes, it can bring together disparate data sources in memory and calculate new aggregates as necessary. It has been recompiled for 64-bit, but does not take special advantage of the hardware. They have a large customer base, but the acquisition by Oracle throws their market traction into doubt.

Applix/Palo: Applix and it’s open source cousin Palo are considered together. Applix is an in-memory implementation of MOLAP. It departs from the traditional MOLAP model in that it holds data in memory, executes queries in memory, and calculates measures in memory. That said, it only scores a single check mark because none of its implementation takes special advantage of being in memory – it is simply a reimplementation of MOLAP. Thus, all the disadvantages of MOLAP (detailed above) still hold. They are a public company, and thus have the advantage that existing on the public market affords. Their customer traction has stagnated over the prior couple of years.

Spotfire: Spotfire does not hold data in memory, execute queries in memory, or build aggregates in memory. Its underlying data store is an Oracle database, and its server uses a BEA J2EE app server. That said, it has a fantastic visualization layer that brings new life to data. It has decent customer traction, but has not made a significant impact in the market.

Hyperroll/TimesTen/SybaseIQ: This group of companies is taken together and represent what Gartner calls “fast query databases”. These tools are simply in memory representations of relational databases. They serve only to speed up query performance and provide basic aggregations. They are not designed as replacements for OLAP analytics. All of these companies are squarely niche vendors who have very few customers and limited market presence.

QlikTech’s Market Strategy and Opportunity

A market for In-Memory Analysis solutions is emerging, and QlikTech is the clear market leader. QlikTech’s market opportunity is to capitalize on this position and use it to drive its continued growth.



If we segment the BI market by two dimensions – the skills of the end user and the functionality that a BI solution provides – we see how today’s solutions are grouped. Excel is a tool designed for untrained or lightly trained users to perform ad hoc data creation and modeling. Users are free to add “records” (rows) and “fields” (columns) at will with out regard to data integrity. They can modify any value at anytime and enter ad hoc formulae that need not be applied consistently. This is tremendously valuable for tasks which require this flexibility. We expect that future versions of Excel will add power user functionality to the traditional Excel model, and extend into the reporting space. Reporting is designed for untrained users to get “their” segment of the data, typically in tabular form. The concept is to provide untrained users with static views onto data. OLAP, by contrast, is designed for highly expert to moderately trained users to gain access to their data and view aggregations of that data. For instance, showing a user total sales by region. Where OLAP is limited is when those aggregates are not pre-built into the cube. SAP BI Accelerator is represented on our market map as extending the OLAP world into more dynamic interactive aggregates. Recall, from above, that SAP BI Accelerator allows queries and aggregates not pre-built into the cube to be generated quickly.

The market opportunity for QlikView is twofold: 1, to bridge the gap between OLAP and Reporting; 2, to grow the market by introducing BI for the mass user. Largely our success to date has been a result of a focus on bridging the gap between OLAP and Reporting. Our customers use QlikView to bring the power of multi dimensional analysis to the untrained user. This bridge allows users the freedom to navigate data across dimensions and while viewing metrics, but QlikView is simple enough for the most basic user to use. QlikView also, however, opens up the possibility of mass deployment of powerful analysis to untrained users. There are, by definition, more untrained users in the world than highly specialized users. By making analysis simple, QlikView is able to grow the market.



SAP

Company Background

SAP was founded in 1972 and is headquartered in Walldorf, Germany. SAP is a public company traded on the New York Stock Exchange under the ticker SAP and on several other exchanges including the Frankfurt stock exchange. It has a market capitalization of US$ 55.9 Billion.

Revenue (Last 12 Months): 11.50B Revenue Per Share (Last 12 Months): 9.3 Qtrly Revenue Growth (yoy): 8.90% Gross Profit (Last 12 Months): 6.63B EBITDA (Last 12 Months): 3.42B Total Cash (Most Recent Quarter): 3.24B

SAP builds and sells a wide range of software, from operational business applications to data integration, to business intelligence. All of these applications are available in industry specific versions targeted to a specific vertical industry segment.

• mySAP ERP application supports analytics, human capital management, financials, procurement and logistics execution, product development and manufacturing, sales and service, and corporate services.

• mySAP PLM application enables organizations to manage the lifecycle of a product, such as initial concept design and engineering, production ramp-up and product change management, and service and maintenance.

• mySAP SCM application enables organizations to manage their supply chain by enabling to synchronize the supply chain, including sourcing, manufacturing, distribution, and fulfillment, as well as to view inventory levels, orders, supplier and customer allocations, forecasts, production plans, and performance indicators.

• mySAP SRM application automates the procurement process through a self-service requisitioning process. The company also provides consulting and education, ramp-up services, hosting, and support for business process outsourcing and custom development.

Overview of In-Memory BI Offering

In late 2004, SAP made a press announcement announcing an “in-memory” solution for business intelligence which was scheduled to be released in May 2005. Specifically they announced SAP NetWeaver would combine “In-Memory computing and advanced search technologies” for an innovative business intelligence solution. A link to the full press release is http://www.sap.com/company/press/press.epx?pageview=print&pressID=3072. The product has subsequently been named SAP BI Accelerator – although the name seems to have changed frequently, and may not be settled at this point.

The product was made available to 30 customers in October, 2005. These were hand selected customers, and the product was not made generally available.



The product was made generally available in May 2006. It is not clear how many customers have signed on from this point on.

Technology. Traditional SAP BI/BW is implemented using traditional OLAP concepts. SAP BW ships a set of pre-defined cubes (called InfoCubes in SAP’s language) which provide analysis of underlying SAP applications. End users query these cubes using SAP BEx (SAP Business Explorer) or another OLAP query front end. This traditional architecture faces the traditional issues: it requires that the analysis is fixed up-front, and is difficult/time-consuming to modify and change in the face of changing business requirements.

The BI Accelerator is a radical departure from the traditional OLAP cube architecture, but is built entirely on top of SAP BW cubes. The BI Accelerator builds indexes of the data in the underlying cubes and, when a query cannot be resolved using a pre-defined cube, it queries the full index. Values which are retrieved from these indexes are then calculated into measures in memory across many multiple processors (much like QlikView would do). These calculations and aggregates are designed to be calculated across multiple machines, so queries are passed to multiple machines and combined at the engine level. The underlying indexing technology is the SAP TREX full-text search/indexing component.

The SAP BI Accelerator has been made available jointly with HP, IBM, and Intel, with Intel being the common thread between them. HP and Intel were first to release it, with how “real” this is, or how much of it is positioning.

Here is a link to a complete demo of the SAP BI Accelerator:

https://www.sdn.sap.com/irj/servlet/prt/portal/prtroot/docs/library/biw/g-i/High%20Performance%20Analytics%20-%20Demo/resources/eventPlayer.htm

Positioning. Technologically, the BI Accelerator is positioned as a fast, easy way to accelerate the query performance of SAP BW. From a business perspective it is positioned as a way to allow freedom in exploring SAP data without being constrained by IT’s pre-determined analysis (read: OLAP cube). Here are the advantages as articulated by SAP:

• Lower total cost of ownership as a result of off-the-shelf hardware, easy installation, and reduced administration

• Flexibility, thanks to on-the-fly aggregation for any query at any time • Scalability, a result of data partitioning and other techniques that optimize the use

of scalable hardware • Performance benefits that include a 10-times to 100-times average-speed increase

and up to 80% faster load times • Maximum return on investment, thanks to increased user adoption of BI and

analytics as well as to productivity advantages gained through better, more timely insights and decisions

Interestingly, this product was featured at SAP’s Sapphire conference in Orlando in May 2006. The product was positioned very similarly to how QlikTech might position QlikView for analyzing SAP data. To quote Shai Agassi during his keynote: “…One of



the biggest issues in a data warehousing is that you have to figure out all the queries you are going to get to build cubes ahead of time. You have to almost think what your users will think, and pre-think to design the architecture of a data warehouse. So what we [SAP] have done with BI accelerator is effectively removed the need to build the cubes…. No more Cubes…. You put the data, stored in the BI accelerator, it stores it in different way than before, but it stores it on-line, immediately accessible, and you don’t need to pre-think of any question that any user will ask…. So No Cubes…”

Strengths

• Positioning. SAP is responding to a valid criticism of it’s OLAP cube approach to analysis, and providing a mechanism for avoiding the pain of the OLAP cube approach.

• Works well with SAP. If you have an SAP-only implementation, with SAP BW, and only want to analyze SAP data, this approach makes sense

• Appliance. Delivering the software as an appliance with Intel/HP/IBM is a good way to address concerns about the feasibility of managing a large server array with complex indexing software. Side note: Google takes this same approach for enterprise text/document search with the Google Search Appliance.

• Scales across machines. SAP is responding to the trend toward highly distributed blade computing by allowing the system to run across machines versus being restricted to a single massive machine. QlikView currently takes the latter approach.

Weaknesses

• Only works with SAP. If a customer requires analysis of data that is stored outside SAP, or not in SAP BW (i.e. cubes) this solution will not work.

• Still requires cubes. The BI Accelerator indexes are built from SAP BW Cubes, thus you must have the cube before you can index it. In short, you don’t achieve Shai Agassi’s vision of “no cubes”.



“Proprietary hardware”/appliance. The appliance approach does not leverage standard off-the-shelf hardware. These machines are single purposed to this task.

• Expensive hardware. The online demo (link above) shows an implementation with a billion records which requires at lease 36 CPUs for usable performance (8 second query times). This would be 9 4-CPU machines. A non trivial amount of hardware. QlikView’s billion record demo uses 8 CPUs.



ORACLE/SIEBEL ANALYTICS

Company Background

Oracle bill itself as the world’s largest enterprise software company. It is publicly traded on the NASDAQ stock exchange under the ticker symbol ORCL. It has a market capitalization of US$ 76.58 Billion.

Revenue (Last 12 Months): 14.38B Revenue Per Share (Last 12 Months): 2.768 Qtrly Revenue Growth (yoy): 25.1% Gross Profit (Last 12 Months): 9.15B EBITDA (Last 12 Months): 5.81B Total Cash (Most Recent Quarter): 7.61B

Today Oracle's Business Intelligence software products are the combination of the products they acquired from Siebel Systems and the previous products which they offered in the market. Oracle completed the acquisition and Siebel Systems on January 31, 2006 and has been working since that point to integrate the two product offerings.

Oracle appears to have an explicit strategy of using acquisitions as a way to grow the presence and software technology market your science. In the past three years Oracle has done no less than 21 separate acquisitions, and boasts of that fact prominently on its website. Acquisitions include: 360Commerce; Context Media, Demantra, G-Log, HotSip, i-flex, Innobase, Net4Call, Oblix, OctetString, PeopleSoft, Portal Software, ProfitLogic, Retek, Siebel, Sleepycat, Telephony@Work, TempoSoft, Thor Technologies, TimesTen, Triple Hop.

In addition to Siebel, the other acquisition that is relevant to Oracle's business intelligence software offering was the acquisition of TimesTen. Oracle completed this acquisition June 20, 2005.

Overview of In-Memory BI Offering

Prior to its acquisition of Siebel (Analytics) and TimesTen Oracle’s business intelligence offering was very standard and did not leverage memory or 64-bit technology. Specifically, Oracle OLAP was a standard cube-based analysis platform and Oracle offered tools for loading data (ETL) and building data warehouses.

Post acquisition, these components still remain, but have been augmented by TimeTen technology and Siebel Anatlytic’s front end. Oracle packages it’s offering in three levels:

• Oracle BI Standard Edition One: “full suite” has all the components of Enterprise Edition, but limited to 2 CPUs. Designed for mid-market

• Oracle BI Standard Edition (SE): Standard OLAP offering, comprised of “old” (i.e. non-Siebel) technology.

• Oracle BI Suite Enterprise Edition (EE): Contains the Standard Edition offering, plus the components previously called Siebel Analytics. These are:



o Oracle Business Intelligence Server: The “server”. More complete technology description below.

o Oracle Business Intelligence Dashboard: The “end user UI” which allows users to manage reports, share reports, and interact with reports on the server.

o Oracle Business Intelligence Answers: The “power user UI” which allows users to create reports and do ad-hoc analysis on the server.

o Oracle Business Intelligence Delivers: The capability to send (i.e. deliver) reports to mobile devices based on alerts and conditions in the underlying data. Example: send me a report on sales when total sales are 20% less than quota.

o Oracle Business Intelligence Disconnected Analytics: same as the Dashboard, but off-line. Only provides static analysis, not real access to the data.

o Oracle Business Intelligence Reporting and Publishing: static reporting o Oracle Business Intelligence Briefing Books: the same as Disconnected

Analytics, but eliminates the capability to drill down.

The Oracle Business Intelligence Server (previously: Siebel Analytics Server) is where the in memory technology that Siebel developed is uses. Siebel Analytics takes the queries that a particular report requires and splits it across any number of underlying data sources and then aggregates the results in memory before returning them to the user. This allows a report to contain data which might exist across many data sources to be visually represented in a graph/chart/table. The major limitation of this approach is that the analytic server is still constrained by the performance of the underlying query engines.

A typical example of this might be a report showing total sales by sales person. This report might contain data from two data sources. First, data comes from the OLAP cubes which are built on a nightly basis. This data source is optimized for this type of query and already contains a cube for total sales by salesperson. The second piece of data comes from the SFA system. This query is for the sales that occurred after the OLAP cube was built (i.e. sales from after last night but prior to “now”). This data source is operational and therefore not optimized for such a query. In constructing this chart, the first query might come back very quickly, but the second one (especially if it were complex) might take minutes or hours.

TimesTen offered an in memory data caching technology useful for speeding access to data stored in transactional databases. This technology is what Gartner would refer to as a “high speed query engine”. It provides application-tier database and transaction management built on a memory-optimized architecture accessed through industry-standard interfaces. Optional data replication and Oracle caching extend the product to allow distribution of in-memory caches to multiple machines or networks.



Summary:

Oracle’s offering is much stronger with the addition of the Siebel technology, and does a much better job of taking advantage of in-memory technology. That said there are some serious limitations to the Siebel Analytics approach, especially when it comes to performance.

Strengths: • Broader than just “OLAP on Oracle databases”. Allows data to come from

anywhere. • Strong user interface and collaboration concepts. • Reasonable performance when all data comes from OLAP sources. • TimesTen: great query performance, especially when build custom applications

on top of an Oracle database

Weaknesses: • Confusing break down of offering into many pieces. • Expensive, especially when taking into consideration how many different things a

given customer needs to buy. • Constrained by the performance of the underlying source systems. • Terrible performance in typical customer deployments. This is because customers

believe the strengths above, but don’t take into account the bullet above. • TimeTen: only fast queries , basic aggregation, no UI. Solves a very limited

problem.



PANORATIO, INC.

Overview:

Privately held spin-off of Siemens headquartered in Munich. With a turnover of 1 million Euros and 10 customers the company must, according to Gartner, attempt to quickly grow the adoption of its proprietary PDI file format to increase usage similar to Adobe's PDF format, possibly through an OEM strategy.

Owners:

SEED / L-EA (ca. 16%) Deutsche Venture Capital Gesellschaft (DVC) (ca. 13%) Polytechnos Venture Partners (ca. 13%) Panoratio Management (ca. 27%) SIEMENS (through Siemens Technology Accelerator GmbH) (ca. 30%)

Locations:

(HQ) Panoratio Database Images, GmbH Theresienstraße 4 D-80333 München Telefon: +49-89-52 03 16-0 Telefax: +49-89-52 03 16-90 (Also office in Stuttgart) Panoratio Database Images, Inc. 631 Howard street Suite 310 San Francisco, CA 94105 +1 415.989.1887 (Also offices in Chicago and New York)

History

Panoratio Database Images, Inc. was established in March, 2003, in Munich, Germany. The technology grew out of work started by Dr. Reimar Hofmann and Dr. Michael Haft at Siemens AG Artificial Intelligence/Machine Learning Laboratory, focused on optimizing the installation and performance of gas turbine engines within Siemens Westinghouse power plants. Sensors placed throughout the power plant generated extremely complex datasets that were impossible to analyze efficiently. Hofmann and Haft solved this problem through development of a pure software solution that reduces the footprint of the dataset without losing any of the detail or complexity. Recognizing the universal value of the solution, Siemens AG, along with Haft and Hofmann, established Panoratio as a separate company. Subsequently, L-Bank/SEED, the venture capital arm of the Landeskreditbank Baden Wuerttemberg, Deutsche Venture Capital and



Polytechnos, Germany’s leading technology venture capital firm collectively funded Panoratio Database Images, GmbH located in Stuttgart and Munich.

Management team

Brett S. Kilpatrick, President & CEO, USA Brett Kilpatrick brings more than twenty years of technology experience to his roles as President and CEO of Panoratio Database Images, Inc. Georg Rybing, CEO & Managing Director EMEA In January, 2005, Georg Rybing took over the management of Panoratio Database Images, EMEA, bringing over 15 years experience as vice president, general manager, sales and marketing director in the IT industry. Stefan Rentsch, CFO Stefan Rentsch joined Panoratio in July 2005 as CFO where he is responsible for finance and investor relations. Dr. Reimar Hofmann, Founder & CTO Reimar Hofmann founded Panoratio in 2003 with Dr. Michael Haft out of work started at Siemens AG Artificial Intelligence/Machine Learning Laboratory. Dr. Michael Haft, Founder & CSO Michael Haft founded Panoratio out of work conducted at Siemens AG Artificial Intelligence/Machine Learning Laboratory with Dr. Reimar Hofmann. As CSO and Vice President Vertical Markets, Michael is responsible for strategic partnerships and positioning Panoratio in vertical markets. Cal Ball, Vice President of Field Operations Sales and Marketing Cal Ball joined Panoratio in 2005 as Vice President of Field Operations responsible for Sales and Marketing within North America. Cal brings more than twelve years of executive experience in sales, marketing and finance operations. Prior to joining Panoratio, Cal served as Vice President of North American Sales for First Virtual Communications. Kevin J. Muerle, Vice President, Professional Services Kevin Muerle brings twenty years of professional services, customer support and operations management experience to Panoratio. As Vice President of Professional Services, Kevin is responsible for Consulting, Customer Support, and IT Operations.

Products:

Database Image Generator The Panoratio Generator is the engine that reduces the footprint of complex data sets to create Panoratio’s revolutionary Portable Database Image (.pdi file). Condenses gigabyte-sized databases into Portable Database Images (PDIs) that are up to 500 times smaller,



whereby all relevant information in the original database is preserved. No pre calculations are made to the dataset.

Database Image Explorer Visualizes Portable Database Images in a way similar to Acrobat Reader. Users can analyze all data intuitively and explicatively – without requiring specialized IT knowledge.

How it works: Panoratio Generator surveys the database and delivers a ”dense” image of the entire dataset in a revolutionary new file format – the .pdi.

• Panoratio Generator will render a complex dataset into a dense image, with density ratios ranging anywhere from 100:1 to 1,000:1. Because the data is not compressed or subjected to cardinality reduction methods, the richness contained in all of the instances and dimensions is retained and made available for driving your analytics and simulation applications.

• The entire dataset resides in memory as the .pdi and no pre-calculations are performed. This means that any and all queries are answered in seconds with only the resident RAM processing power necessary to perform the calculations.

• The Panoratio .pdi delivers to your customer the ability to query all of their data quickly, iteratively and across a virtually unlimited number of instances and dimensions.

• The .pdi is small enough to distribute via download or email, allowing anyone in the organization immediate access to all of the data in a secure, portable format.

• Because users are working independently, running their analyses against the .pdi, your production database remains untouched and unburdened.



Key Customers

KarstadtQuelle: Germany’s leading mail-order company has used Panoratio technology since 2002, allowing all of their marketing decision makers the power to access to their database of over 3.5 million customers in order to make timely, accurate, strategically sound decisions.

Siemens Westinghouse Power Generation (Siemens PG): Siemens PG, a leading enterprise worldwide in power generation, uses Panoratio technology to break through the complexity of identifying the optimal operating states of a gas turbine as a function of operating parameters to implement the optimal operating mode.

NDC Health: An international company that provides information solutions to the healthcare industry needed to investigate and distribute 2 years of prescription history data for about 7 million patients in Germany(about 90 million prescriptions). Panoratio’s technology transformed their 17G database of prescription data into a .pdi of just 280 mb, distributed to product managers monthly. Now, they can provide critical analysis of temporal sequences and co-therapies for patients on-the-spot.

Geographic penetration and coverage

It’s hard to get valid information. Coverage mainly in Germany and United States where they have sales offices. They don’t seem to have developed a partner network. Given that the company employs roughly 20 it is assumed that their international coverage isn’t enormous.

Summary

Panoratio today looks very much like QlikTech in 1998 – a small, technology lead company with few customers. It has an approach to BI which is more flexible than its traditional OLAP competitors and which makes use of the increasing performance of hardware.

Strengths: • In-memory BI solution (fast and flexible approach) • Gets rid of redundancy • Short implementation times • Off-line capabilities for portable analysis • Intuitive and requires little end user training (typically one day) • Fast response times to queries to the entire data set • Seem to be well funded for further expansion

Weaknesses: • Not very mature product line • Off-line capabilities only (no client-server solution) • Technical approach/focus, solution for complex databases • Value proposition focuses on technical solutions, how small the pdi file is and

how fast queries are compared to SQL queries directly to the database



• Does not seem to have the knowledge or insight of the real business needs for BI • Few and probably demanding customers (like Tetra Pak for QlikTech in early

days) • With annual revenue around 1 million euros, about 10 customers, and limited

channel partnerships and marketing capabilities, Panoratio will find it hard to emerge from its highly specialized niche market and rather departmental solutions (Andreas Bitterer, Gartner)



SPOTFIRE

Company Overview

Company Type: Private Company Website: www.spotfire.com Number of Employees: 150 Year Founded: 1996 Annual Revenues (Approx.) ($mm) 15.9

Spotfire, Inc. develops and markets visual data analysis applications to life science, semiconductor, and consumer goods industries. The company’s product DecisionSite, enables user to perform data filtration, expression summarization, structure searching, activity profiling, and pattern detection. Its clientele includes Bayer, Abbott Laboratories, Biogen, Eli Lilly, Novartis, Pfizer, and Sepracor. The company has strategic alliances with DeltaSoft, Inc., Intrasphere, Naxos Solutions, CAS, ESRI, IBM, and Oracle. Spotfire was founded in 1996 and is headquartered in Somerville, Massachusetts with additional offices in Goteborg, Sweden; and Tokyo, Japan.

Primary Office Location: 212 Elm Street | Somerville, MA | 02144 | United States; Phone: 617-702-1600 Fax: 617-702-1700

Current and Pending Investors: Atlas Venture LLP, Casdin Capital Partners, Chevron Technology Ventures L.L.C., Cooper Hill Partners, L.L.C., In-Q-Tel, Inc., Innovationskapital (Staffan Ingeborn), SEB Företagsinvest, Sprout Group (Philippe Chambon), TowerBrook Capital Partners, L.P.

Prior Investors: Chalmers Innovation AB, Pequot Capital Management Inc., Volvo Technology Transfer AB

Technology: Spotfire is positioned as a visualization engine. It is not clear how much “query” and “calculation” they do. In addition, they do not talk much about their utilization of memory, multi-processor platforms, or 64-bit technology. One can assume that they are weak here. It is likely that they are very much a user-interface technology, and not a serious player in the in-memory analysis market.

In addition, Spotfire requires an underlying Oracle database and a BEA application server. Based on the fact that the data is held in a traditional relational database, it is likely that volume of data available to Spotfire is quite small. Also, this is further indication that little, if anything, is done in memory. Spotfire can visualize small amounts of data well – but little more than that.

That said, they have just released their most recent version of their product, Spotfire DXP, and it has clearly taken some cues from QlikView.

Demos of the prior version of Spotfire can be found here: https://ondemand.spotfire.com/Citrix/MetaFrame/site/default.aspx



PANORAMA SOFTWARE LTD.

About Panorama

Panorama Software, the company that sold its leading OLAP technology to Microsoft, helps global organizations unlock the hidden value of their information assets to improve business performance and results through SMART, SIMPLE and COMPLETE BI solutions. Panorama takes advantage of the Microsoft (NasdaqNM:MSFT) and SAP (NYSE:SAP) platforms through smart, simple and complete business intelligence and corporate performance management solutions.

Panorama, a leading innovator of business intelligence solutions, supports customers worldwide in industries such as financial services, manufacturing, retail, healthcare, telecommunications and life sciences. Panorama has more than 250 partners in 30 countries, and maintains offices throughout North America, EMEA and Asia.

History

Panorama was founded in 1993 and released its first OLAP product in 1995. The Panorama OLAP technology was acquired by Microsoft Corporation in 1996. Since 1997 Panorama has been focusing on providing comprehensive BI solutions that embed the powerful Microsoft OLAP engine and exploit it to the fullest degree.

Hundreds (approx 400) of customers worldwide enjoy Panorama’s solutions and rely on them for their daily decision making processes.

• 1993 Original Panorama Software Systems business founded • 1996 OLAP technology sold to Microsoft • 1997 Development of 2nd generation OLAP with NovaView® • 2001 Launched 3rd generation Web-based Panorama NovaView® product line • 2003 Panorama Software Ltd. expands global operations opening offices in New

York, Toronto, London and Barbados • 2004 Panorama launches Panorama 4.0 Integrated Business Intelligence Solutions • Microsoft Technology Centers worldwide feature Panorama Software's Business

Intelligence Solutions • 2005 Panorama Software releases beta of Panorama 4.5 • Panorama Software supports Microsoft's new advanced Business Scorecards

Application • Panorama Software releases candidate 4.5 supports .NET and SQL Server 2005

integration. Panorama Software releases Panorama NovaView® 5. Panorama Software releases Panorama NovaView® Scorecards. 2006 Panorama Software receives SAP Enterprise Portal Certification

• Panorama Software unveils the first complete and integrated business intelligence solutions for SAP



Owners

It’s hard to find information about the owner community but we know that Intel Capital has been investing in the company.

Intel says the following about the company:

The Panorama NovaView BI solution is a multi-layered software system, with a middle-tier server responsible for connecting to the data sources, running and managing queries on the data sources, and maintaining performance, scalability, caching, and security. Panorama NovaView offers multiple platform solutions – connecting either directly to SAP BW, or via Microsoft SQL Server to non-SAP databases. Panorama NovaView enables end users to access the browser and open a personalized application, which integrates all business information required by the user in a single interface. Users can create and track personal and organizational objectives, display, analyze and share business data, collaborate with colleagues, and more. NovaView offers a complete set of drill-through, slicing and dicing, and analysis tools that enable users to deconstruct trends and alerts, and isolate specific instances and activities. Panorama NovaView is highly proactive, providing numerous configurable notice and alert options, which generate information and empower decision-makers to take initiative and respond immediately to emerging issues and concerns. NovaView also enables the integration of BI into business processes and workflow scenarios, where intelligent decisions can be automated in order to enable decision-makers to focus on quality-impact analysis and implementation. The system is highly dynamic, enabling top-down analysis that allows the user to access and manipulate source data. Flexible and powerful, Panorama NovaView has multiple security layers, insuring that enterprise users can only access and utilize appropriate information. NovaView provides its customers with a user-friendly system which delivers the right information at the right time, amplifying business impact consistently. With rapid ROI, low TCO, fast deployment and complete integration with existing infrastructure, Panorama NovaView is an invaluable business tool that provides a vital an edge in a highly competitive global economy.

Locations:

Worldwide Panorama Software Ltd. 164 Eglinton Avenue East, Suite #400 Toronto, ON M4P 1G4 Canada Phone: +1.416.545.0990 United States Panorama Software Inc. Route 17 North, Building 201, Suite 400 Rutherford, NJ 07070 Phone: +1.201.804.0144 Europe



Panorama Software Europe Ltd. Berkeley Square House, 2nd Floor Berkeley Square, London W1J 6BD Phone: +44.207.887.6300 Israel, Asia-Pacific Panorama Software Israel Ltd. 6, Raoul Wallenberg St. 5th Floor Ramat Hachayal Tel Aviv, 69719 Israel Phone: +972.3.645.9777 Sales offices in: New York, Paris, and Munich

Management team

Eynav Azarya, CEO: As Chief Executive Officer for Panorama Software, Eynav is responsible for the strategic vision and direction of the organization.

Mark Corsetti, Senior Sales Contact : Mark Corsetti has over 20 years of Sales, Marketing and Services roles within technology and software solution companies. As Vice-President Worldwide Sales Mark will lead the continued growth of the Western European and America’s marketplaces.

Lee Ho, Vice President: Lee has overall responsibility for strategy and execution of product marketing, vertical market development, branding, global awarness, channel marketing, and marketing communications worldwide.

Rony Ross, Chairman/Founder: Rony Ross is the Founder, Executive Chairman and Chief Technology Officer of Panorama Software Ltd.

Products

Panorama NovaView is a multi-layer business intelligence platform that incorporates an Intelligence layer for content management and information delivery and a Presentation layer for accessing and analyzing data.

Panorama's architecture supports multi-level permissions for content viewing, targeting and scheduling delivery of selected data to specific end-users.

Panorama NovaView offers:

• A complete set of drill-through, slicing and dicing, and analysis tools • The ability to deconstruct trends and alerts, and isolate specific instances and

activities • A highly dynamic solution, enabling top-down analysis that allows users to access

and manipulate source data



• Multiple security layers, insuring that enterprise users can only access and utilize appropriate information

Intelligence Layer

The Intelligence Layer contains the NovaView Intelligence Server, Panorama's high performance Web application server as well as content management and information delivery services and an XML repository. The NovaView Intelligence Server stores and caches data and manages access and delivery of information to end-users. The operational data source can be any database, such as Oracle®, Sybase®, DB2® and SQL Server™. The data warehouse layer contains the Microsoft OLAP engine and cubes as well as a data warehouse/data mart.

Presentation Layer

The Presentation Layer provides various Web-based and Win32 client interfaces that allow users to access and perform analytics on data stored throughout your organization. Clients can communicate directly with the OLAP cubes (which connect to any transaction system or database) or, in the case of the NovaView Web Access and NovaView for Microsoft Excel clients, through the NovaView Intelligence Server.

The NovaView SDK can be used to seamlessly integrate Panorama's multi-dimensional analytic functionality into third-party applications, corporate intranets and portal applications.

Panorama offers a variety of NovaView products:

NovaView Server The Panorama NovaView Server is the key component of the NovaView solution. The Server provides a middle tier between clients and the Microsoft SQL Analysis Server/SAP BW, performing most of the calculations and caching associated with the OLAP queries sent through the Panorama client applications. All the business logic, metadata, and Smart capabilities are processes in the Server.



NovaView Analytics NovaView Analytics enable decision-makers of all levels to analyze and report against dynamic organizational data, providing rapid inside into ongoing business performance. With a variety of drill-down and drill-through options, NovaView Analytics provide users with tools to investigate performance and analyze data to isolate specific organizational dynamics, trends, and events, pinpointing the triggers and sources of significant activities.

They also offer separate products for reporting, dashboards, scorecards, business modeling, and a SDK for integrated and customized solutions.

Key Customers:

Barclays Bank United Kingdom Financial Services Cadbury Trebor Bassett United States Manufacturing EMI Music United Kingdom Media France Telecom EMEA Services Greyhound Lines Inc. United States Services Janssen-Cilag Latin Amaerica Manufacturing Jelly Belly Candy Company United States Manufacturing Manpower Israel Services Mobiltel EEMEA Services Motorola Israel Manufacturing MTV United Kingdom Media National City Bank United States Financial Services Nestle EEMEA Manufacturing Royal Bank of Scotland United Kingdom Financial Services Texas Instruments United States Manufacturing ToysRus Israel Retail Visa Israel Financial Services


Panorama focuses on five main areas which are Retail, Financial Services, Manufacturing, Health Care, and Media. As of today they have approximately 400 customers. Panorama also has more than 250 partners in 30 countries, and maintains offices throughout North America, EMEA and Asia.

Summary

Panorama’s NovaView does not perform in-memory analytics in the traditional sense. It only caches SQL like statements for end user queries which can be shared between large user groups for faster response times.

Panorama is what we would call a traditional OLAP vendor and their technology does not differ much from most of our competitors. Their solution only works with Microsoft SQL Server and SAP BW which narrows their market.



It’s a mature company with mature products and a good knowledge of the information pain many companies suffer from. They are MS Gold Certified Partners and SAP Certified Partners. Panorama is also known for being the company who in 1996 sold their OLAP technology to Microsoft. This technology is used in SQL server analysis services.

Strengths: • SAP certified • Strong offering for SAP customers • Visually appealing dashboards and scorecards • What if and write back capabilities • Strong drill through capabilities for more detailed analysis caching query

statements for reuse among users • Integration with MSFT SharePoint

Weaknesses: • Traditional OLAP (not market leader) • In Gartner’s MQ but in lower left (no challenger, lack of vision) • Solutions limited for SQL Server and SAP BW only • Microsoft and SAP are developing and selling their own BI solutions



MICROSTRATEGY

Overview

MicroStrategy claims to excel in all five types of BI on the market (Scorecards and Dashboards, enterprise reporting, OLAP, Advanced and Predictive Analysis and Alerts and Proactive Notification) instead of focusing on one method.

Having had financial problems around the year 2000 followed by a time of aggressive pricing, the company now seems to stand on firmer ground. MicroStrategy is climbing the “Slope of Enlightenment” according to Gartner (Hype Cycle for Business Intelligence and Corporate Performance Management, July 2006).

Owners

Traded on NASDAQ under the MSTR ticker: 26% of shares held by all insider and 5% owners; 69% of shares held by institutional & mutual fund owners; 92% of float held by institutional & mutual fund owners.

Locations

MicroStrategy has more than 300 systems integrators and application development and platform partners including IBM, PeopleSoft, Sun, HP and Teradata. Sales offices are located in major cities throughout the United States, Canada, Europe, South America and the Asia Pacific region. MicroStrategy has 10 offices in the US, 9 EMEA offices, 32 EMEA distributors and 44 distributors in Asia. The company has around 1000 employees.�

Company HQ: 1861 International Drive McLean, VA 22102 Tel: 703.848.8600

History

Founded in 1989 by Michael Saylor, MicroStrategy became infamous for an accounting debacle in 2000. After having the balance sheet sorted out along with an extensive cut in staff and elimination of non-core business such as Strategy.com, the CRM applications division, hosting & ASP services, and systems integration, the company became profitable in 2003. MicroStrategy is now, according to Gartner (Kurt Schlegel and Bill Hostmann), climbing the Slope of Enlightenment with a well-designed version 8 and a high user acceptance.

Management team

Michael J. Saylor. Chairman of the Board, President and Chief Executive Officer. Sanju K. Bansal. Vice Chairman of the Board, Executive Vice President, Chief Operating Officer and Secretary Arthur S. Locke III. Vice President, Finance and Chief Financial Officer



Jonathan F. Klein, Vice President, Law and General Counsel Jeffrey A. Bedell, Vice President, Technology and Chief Technology Officer Eduardo S. Sanchez, Vice President, Worldwide Sales Adam M. McDonald, Vice President, Worldwide Services Paul Zolfaghari , Vice President, Worldwide Business Affairs

Board members:

Michael J. Saylor. Chairman of the Board, President and Chief Executive Officer. Sanju K. Bansal. Vice Chairman of the Board, Executive Vice President, Chief Operating Officer and Secretary Matthew W. Calkins. Founder, President and CEO, Appian Corporation Robert H. Epstein. President and CEO, Takeda Lace USA, Inc. F. David Fowler. David W. LaRue. Jarrod M. Patten. Founder, President and CEO, Real Estate Resource Group, LLC Carl J. Rickertsen. Managing Partner, Pine Creek Partners Thomas P. Spahr. President, Libra Ventures, LLC

Technology and products

MicroStrategy 8 platform • MicroStrategy Intelligence Server: The industry's first and only BI server to

provide all 5 styles of BI on a single, unified platform using a services-oriented architecture.

• MicroStrategy Narrowcast Server: A proactive information delivery server that distributes personalized business information to users via email, printers, file services, SMS and mobile devices.

Service Modules • MicroStrategy OLAP Services. An extension of MicroStrategy Intelligence

Server that allows MicroStrategy Desktop, Office and Web users to manipulate Intelligent Cubes™. Combines MOLAP with ROLAP analysis.

• MicroStrategy Report Services. The enterprise reporting engine of the MicroStrategy BI platform that delivers the entire range of enterprise reports.

• MicroStrategy Data Mining Services. A fully integrated component that delivers data mining predictive models to all users.

• MicroStrategy SAP® Services. SAP certified NetWeaver™ integration with SAP BW, SAP WAS & SAP Enterprise Portal.

User Interfaces • MicroStrategy Desktop. The BI software component that provides integrated

query and reporting, powerful analytics and decision support workflow on the personal computing desktop.

• MicroStrategy Office. An add-in that gives business users open and straightforward access to the full functionality of the MicroStrategy platform -- all from familiar Microsoft Office applications.



• MicroStrategy Web. A highly interactive, easy to use interface for monitoring, reporting and analysis.

Development Tools • MicroStrategy Administrator. A suite of tools that provide the most

comprehensive systems management environment for business intelligence. • MicroStrategy Architect. A rapid development tool that maps the physical

structure of the database into a logical business model & stores it in a centralized metadata repository.

• MicroStrategy BI Developer Kit. A CPM starter kit with application modules & development tools that lets business users immediately start reporting, analyzing, & monitoring corporate performance.

• MicroStrategy SDK. A comprehensive development environment that enables quick & easy customization & integration of the MicroStrategy platform into any application on any platform.

MicroStategy also provides the following “Starter Kits”

• Customer Analysis • Financial Reporting Analysis • Human Resources Analysis • Sales Force Analysis • Sales & Distribution Analysis • Web Traffic Analysis

The MicroStrategy 8 platform is based on version 7, which was a re-write of earlier versions. It contains both JSP and ASP.NET front ends and integrates well into Office.

MicroStrategy concentrates on developing all its software in-house and does not acquire technology through acquisitions. It still features traditional OLAP technology and does not venture into in-memory analysis. The product requires a very well-defined underlying data structure in the form of a star schema.

Future

2005/2006 marketing targets (According to MicroStrategy): • Our historical base of corporate technology buyers and departmental technology

buyers in Global 2000 enterprises • Corporate and departmental technology buyers in mid-sized enterprises, with

annual revenues between $250 million and $1 billion • Government technology buyers and the vendors to the government community • Independent software vendors who want to embed analytical tools in their

solutions • System integrators who have technology relationships with the largest 2,500

enterprises, governments and information intensive businesses.



2005/2006 Technology Strategy. (According to MicroStrategy) The MicroStrategy technology strategy is focused on expanding its support for large information stores, enhancing analysis and segmentation capabilities, strengthening its personalization technology and enhancing report delivery and alerting functionality to all commonly used devices. The company continues to enhance its technology for use with a broad range of operating systems and databases to enable customers to leverage their existing technology investments to achieve faster query times with fewer required resources. In addition, MicroStrategy continues to develop its platform for easy integration with a wide spectrum of ERP systems.

2005/2006 Sales Strategy. (According to MicroStrategy) The sales strategy focuses on direct sales through their dedicated sales force and relationships with indirect channel partners in order to increase market share in both domestic and international markets. MicroStrategy also seeks to increase sales to the existing base of customers by offering a range of software and services utilizing the integrated business intelligence platform.


The MicroStrategy market penetration has a strong North American bias, especially in the retail industry. Generally, the MicroStrategy 8 platform seems to be targeted towards larger enterprises with the aim of becoming organization standard BI software.

Revenue

MicroStrategy grew 13.7% in 2005, 2% less than previous year. The company is reported to hold about 7% of the BI market, about the same percentage as Business Objects.

The company has revenue of $268.7 million. Large resources have been spent on buying up own stock. MicroStrategy does not offer predictions on sales and earnings, making it a difficult organization to penetrate.

Summary

MicroStrategy is, once again, back in business. The version 8 suite is something of a Jack-of-all-trades, allowing it to reach a wide range of customers but running the risk of not being excellent at anything. By providing a wide range of methods, and with good educational material and support, the suite is positioned to fit any large organization and fill all the BI needs of the customer. The company is rated as a leader by Gartner and is profitable. The technology is, however, based on OLAP cubes and static reports and offers little beyond the nearest competition.

Marketing will be targeted towards existing install base, mid-size organizations ($250 million and $1 billion), government and OEM-type operations. Development will be focused on facilitating larger sites and more users. Sales will be driven through up-sales, education and services.



Strengths

• Integrates with Office • Web-based SOA (MicroStrategy Intelligence Server) architecture with zero-

footprint clients. • Focus on enterprise-wide solutions by interacting with major data sources and

other cubes.

Weaknesses

• Demands star schema. • Difficult configuration.



APPLIX

About Applix:

Like many companies in the BI space, Applix has had trouble deciding whether it sells applications, BI system software, or both. Its core technology is TM1, a memory-centric MOLAP offering. Traditional MOLAP products reside on the horns of a nasty dilemma: they rely on pre-calculation to give good performance, but that causes significant explosion in database size. By working with data in memory Applix avoids needing to store pre-calculations and rather executes all queries on the fly.

Applix has an open source competitor (read: copy) called "Palo" (http://www.palo.net/). Of course, the product is freely available for install/download. The Palo site is quite good (in the sense that its "open"), and I would suggest reading the evaluation guide they publish. It is a good walk through/demo of what they offer. What is amazing is how similar it is to Applix.

What is Applix

In short, Applix is very different from QlikView:

Applix/Palo QlikView

They are a tool for building data capture and display tools.

We are a tool for visualizing, querying and aggregating data (and "information").

They deal with small numbers of very hierarchical dimensions with relatively small numbers of unique dimensional members.

We deal with large numbers of non-hierarchical (or, perhaps arbitrarily-hierarchical) dimensions with large numbers of repeating members.

They are well suited to budgeting/planning applications where users need to see small amounts of data, aggregated across a number of dimensions, and then add/change data in certain points in the hierarchy.

We are well suited for gathering large amounts of data, and allowing users to "surf" the dimensions while calculating measures in real-time -- typically an optimization or query problem (where are sales low?).

They are numerical/text (spreadsheet). We are visual and interactive.

They don't provide an easy mechanism for arbitrary queries.

We allow click to query.

They lock the dimensions at design time (although it is quiet easy to add them at design time) and lock the user into a given set of dimensions, hierarchies, and drill downs.

We have no fixed dimensions, and unconstrained drill down.



Strengths

• Provides write-back capability. More importantly it allows the user to enter a value at some point in a hierarchy, and have it spread back to other nodes in that hierarchy. So, for instance, you can enter sales in North America, and have it distribute that total back to the States using some calculation (even split, split proportionally, hold one state constant, etc.). Worth noting that it seems like Applix is more advanced here than Palo.

• Fast query performance due to optimized storage, multidimensional indexing and caching.

• Smaller on-disk size of data compared to data stored in relational database due to compression techniques.

• Automated computation of higher level aggregates of the data. • It is very compact for low dimension data sets. • Array model provides natural indexing • Excel based UI model makes for very spreadsheet type experience. • Security at cell level for Applix (not Palo) allows admin to control who enters

data where at the cell level.

Weaknesses

• Charting capability is limited to what Excel can do, which is only very basic line/bar charts. No heat chart, color coding, multi-dimensional, etc. QlikView type charts.

• Any cube is limited to 256 dimensions, and practical limitation is much lower than that.

• Users who don't have Microsoft Excel (or are unwilling to use Microsoft products) are quite limited in their approach. Also the Excel interface puts a premium on "data"/"numbers" instead of text, information and aggregations/measures.

• Serious limits to the amount of data that can be analyzed. Its clear that with 64-bit technology Applix can analyze "more" data, but because the underlying model is a traditional OLAP cube, with more dimensions and measures the amount of data they can handle will be severely limited. In addition, there is an exponential limit to the capability to respond quickly to queries (and updates) as the data, dimensions, and measures increase. Just estimating here, but Applix could handle hundreds of thousands or one/two million records at the high end; QlikView can handle hundreds of millions and billions at the high end.

• The processing step (data load) can be quite lengthy, especially on large data volumes. This is because the tool needs to populate all the measures across all the dimensions. Again, because it is in memory Applix has an advantage here relative to ROLAP tools. But, with large amounts of data or large numbers of dimensions/measures will cause a significant slow down.

• These tools will suffer with data that in which some dimensions have very high cardinality (i.e., millions of members).



Summary

Applix/Palo solve a very different problem than QlikView. They are optimized for write-back to a small number of run-time fixed dimensions. They suffer with large numbers of dimensions/measures, or with large amounts of data. QlikView's strength is data analysis with infinite dimensions/measures, more flexibility, faster/easier querying, and highly interactive visual charts.



INFORMATION BUILDERS

Overview

Information Builders (IBI) is a privately held provider of enterprise business intelligence and reporting software. The company also provides consulting and other services. The company is headquartered in New York City, New York

Information Builders offers software and services that are used in the design of enterprise reporting and decision support systems, data warehouses, cross-platform application development, and integrated application solutions. Their software solutions all share a common middleware architecture for enterprise data access.

The company’s competitive offerings include its flagship WebFOCUS product line, which offers capabilities such as enterprise reporting, ad hoc query and OLAP analysis, information delivery and management, data access/management and ETL, portal integration, development tools, administration and security, and closed-loop business intelligence. The iWay Software subsidiary provides solutions for the deployment and risk management of a range of integration projects, including composite applications, web services deployments, business-to-business interactions using XML and EDI, ETL, data integration, and enterprise information integration initiatives. In addition, the company’s FOCUS systems provide application development and reporting tools for host-based reporting. Information Builders’ solutions are provided to a range of businesses and organizations across the banking and financial services, energy and utilities, government, healthcare and pharmaceutical, higher education, insurance, manufacturing, and telecommunications industries.

(IBI’s own figures) $ 337 million turnover $ 300 million revenue (2003) > 1,750 employees > 12,000 customers

Owners

IBI is privately owned and no information was found on the owner structure.

Locations

Information Builders Corporate Headquarters Two Penn Plaza, New York, NY 10121-2898 Main Phone: +1(212) 736-4433



Sales offices in the US, Mexico, Canada, Europe and Australia. Sales representatives only in other countries. Consulting offices in the US only.

History

Founded in 1975 by Gerald D. Cohen (President and CEO), Peter Mittelman (V.P., FOCUS Division) and Martin B. Slagowitz (V.P., Consulting) and still privately owned, IBI has provided BI to Fortune 100 companies for three decades.

Management team

Company Founders Gerald D. Cohen, President and CEO Peter Mittelman, V.P., FOCUS Division Martin B. Slagowitz, V.P., Consulting

Corporate Executives Timothy Benthall, V.P., Systems and Communications Michael Corcoran, V.P., Chief Communications Officer David Kemler, V.P., Sales & Marketing Harry Lerner, V.P., Chief Financial Officer Wesley Thompson, V.P., Channels Dennis McLaughlin, V.P., iWay Software Sales David Sandel, V.P., Business Intelligence Products Group John G. Senor, President, iWay Software

Technology and products

The FOCUS concept with FOCUS (mainframe and UNIX versions) and the web-enabling version WebFOCUS 7 works through creating a lookup table that maps against the underlying data. In essence, this is cube technology in another technical solution. The lookup table, the Master reference files, are cached on the server, making queries run reasonably fast. This also makes the product sensitive to the amount of available RAM and the speed thereof.

By providing adapters towards different data sources, IBI enables reading from, and comparative analysis from different data sources. These adapters are developed by the subsidiary iWay Software. iWay essentially provides an Enterprise Service Bus, a middleware layer that acts as a connection between the underlying data sources and the reports and the adapters that enable the connection to these data sources. Inform Applications, a part of IBI provides reports tailored for RMIS (risk management information solutions), Claims, OSHA (Occupation safety and health) and the Insurance industry.



The WebFOCUS 7 component ReportCaster provides scheduling, distribution and storage of information, mainly as WebFOCUS reports, but also as third-party content.

Implementation is provided by Information Builders Consulting services or by any of the IBI partners.

Key Customers

• Banco Bilbao Vizcaya Argentiara • Bombardier Capital • California Department of Health Services • FEMA • Ford Motor Company • John Hopkins University • Lloyd’s of London • Macy’s • PWC • Sony • Google


From the information available, mainly US, but with some EMEA implementations and a handful in Asia. With a fairly large NA presence with representation in many US states and neighboring countries and sales representation in many western European countries IBI has targeted large organizations and seems to focus on enterprise-wide deals.



Revenues (2003 figures):

• WebFOCUS suite and FOCUS: 61% • iWay Software (integration products): 14% • Consulting: 25 %

Customer breakdown by top five industries (percentage of revenue):

• Finance, banking, and securities – 16.71% • Insurance – 14.92% • Manufacturing (including aerospace and automotive) – 12.64% • Government – 10.06% • Telecommunications – 9.93%

Summary

IBI has taken a server with installed client tools and made the tools web-based/browser based. Moving from FOCUS to WebFOCUS is something that IBI promotes. By selling software with targeted applications and providing implementation services, IBI is a start-to-finish operation. Alliances with BEA and IBM makes IBI strong on the SOA field. They are strong in North America and in financial and insurance as well as public services. Strong points are web-based access, balanced scorecards and integration with other technologies such as platforms, portals and security systems.

Strengths:

• Web-based (WebFOCUS) • Full service provider • Uses trendy technology (SOA, AJAX) • Strong customer base

Weaknesses:

• Master reference file technology • Difficult configuration

(http://www.bizintelligencepipeline.com/showArticle.jhtml?articleID=17120413) • Weak ad-hoc query capabilities



OTHER VENDORS

• TimesTen (now owned by Oracle): TimesTen is the quintessential “in-memory DBMS.” It’s a fairly full relational DBMS, but if you want to store what’s in memory to disk it has to be handed off to a conventional DBMS. Historically, that has usually been MySQL or Oracle. TimesTen’s biggest market penetration has been in financial trading.

• Solid Information Technology’s BoostEngine: Solid is a Finnish company (or was — it’s pretty American now) specializing in embedded DBMS sold mainly for telecommunication uses. Big OEM customers include several well-known telecom equipment manufacturers and HP (for OpenView). “Embedded” often means no DBA, no monitor, no keyboard — they box manufacturer installs it and there it stays for the life of the product. Solid has to offer strong replication capabilities, since its products are often used in highly distributed (e.g., multiblade, multibox) environments. So it’s taken the next step and exploited the replication by allowing customers to use some instances of the product disklessly.

• Event-stream products from Streambase and Progress: The canonical application for event-stream products is automating financial trading decisions based on the flow of market information. Mike Stonebraker, the brains behind Streambase, has recently popularized the idea; Progress bought Apama, who actually have been in the business longer. These applications require even more speed than the financial trading apps that TimesTen handles, and they discard most of the information they look at. In-memory is the only way to go.

• Progress’s ObjectStore: ObjectStore comes from the company Object Design, which merged into Excelon, which was acquired by Progress. It’s really a toolkit for building DBMS and similar systems, which is why it’s at various times been marketed as an OODBMS and an XML DBMS, without a lot of success either way. But there have been a few sterling apps built in ObjectStore even so, including a key part of the Amazon bookstore. Despite this limited market success, a significant fraction of Progress’s best engineering talent has moved over to the Real-Time Division to focus on ObjectStore and other memory-centric products. The memory-centric aspect of ObjectStore is this: ObjectStore’s big virtue is that it gets objects from disk to memory and vice-versa very efficiently, then distributes and caches them around a network as needed. This was originally invented for client/server processing, but works fine in a multi-server thin client setup as well. And object processing, of course, relies on a whole lot of pointers. And pointer-chasing is pretty much the worst way to deal with the disk speed barrier, unless you do it in main memory.



64-BIT MARKET PENETRATION 64-bit hardware technology has been gaining market acceptance for the previous 2 years, and today should be considered mainstream. According to Intel, in today’s database market 80% of hardware sales volume is 32-bit, but 80% of spending is on 64-bit.

According to IDC reporting on Q4 2005 (Feb 22, 2006): “The rapid transformation of the x86 marketplace into a segment that is 64-bit enabled continued with x86-64 based systems accounting for 78.8% of all x86 server spending, with factory revenue for x86-64 systems more than doubling year over year.”

From Gartner Research, Nov 16, 2005:

• Prediction: By year-end 2007, Intel will be actively developing a new 64-bit processor as a follow-up to Itanium and Xeon

• Key Findings: The server market is moving toward processor architectures that use 64-bit computing, but Intel’s current processor strategy does not properly line up with this movement. Its Pentium and Xeon lines have been “extended” to support 64-bit computing, but they are not 64-bit processor architectures. By contrast, Intel’s Itanium is a 64-bit architecture, but price and performance issues with 32-bit x86 software have hampered Itanium’s penetration into the broader x86 market. As a result, by year-end 2007, Gartner expects Intel to be actively developing a new 64-bit processor product as a follow-up to Itanium and Xeon. This follow-up product will directly replace Pentium and Xeon, and it will have the effect of further limiting Itanium’s future market share.

• Market Implications: Servers based on the x86 platform represent the bulk of shipments and revenue in today’s server market. Itanium has tried to penetrate this market, but has had limited success. At the same time, Intel’s x86 Pentium and Xeon processors are aging designs founded on a 32-bit architecture. Intel, with its majority share of the market at risk, must develop a followup product to maintain its dominant position — even if that product adversely affects sales of its other server processor, Itanium.

• Recommendations: Consider it safe to invest in x86 software solutions beyond a three-year time frame. Intel will not repeat the mistake it made with Itanium by offering a new processor architecture with poor appeal to legacy software. IT organizations should also keep a careful eye on the areas where they invest in Itanium servers. The high-end and midrange servers are preferable. But trying to use Itanium servers as a broad replacement for x86 servers running Windows or Linux is risky, because the Itanium software ecosystem will trail behind that of the x86.



Microsoft SQL Server Support for 64-bit

Product Name Architecture Processors OS Requirement

SQL Server 2005 for Itanium-based Systems

EPIC Itanium 2 Windows Server 2003 for Itanium-based Systems

SQL Server 2005 for x64 Editions

x86-64 AMD Opteron; AMD Athlon 64; Intel 64-Bit Xeon; Intel Pentium with EM64T

Windows Server 2003 x64 Editions

The Microsoft ODBC Driver Manager and the Microsoft SQL Server ODBC Driver have been updated to support 64-bit Windows. Windows Server 2003, 64-bit was the first Microsoft operating system to support the 64-bit ODBC libraries. The ODBC headers and libraries shipped with MDAC 2.7 SDK contain changes that will allow programmers to easily write code for the new 64 bit platforms

MySQL

MySQL supports AMD64 and Intel64T. The MyODBC Driver does not appear to support any 64bit platforms: http://dev.mysql.com/doc/refman/5.0/en/myodbc-versions.html

DB2

The DB2 Universal Database V8.2.2 supports x64 Windows on AMD64 and Intel EM64T, as well as Itanium. DB2 9 supports all Intel and AMD processors capable of running the supported windows operating systems (32-bit and 64-bit). I was unable to find anything on ODBC Drivers for DB2

Oracle database

The Oracle database supports 64-bit Windows Server 2003 and XP 2003. Oracle was the first vendor to make database software publicly available for the 64-bit Windows Itanium platform in December, 2000. Oracle9i Database Release 2 was the first production database supported on 64-bit Windows.

Current version platform support includes Itanium and x64:

Oracle Database 10g (10.2) Release 2 Supported Windows Operating Systems

32-bit Database Server and Client 32-bit Windows

32-bit Client 64-bit Windows x64

64-bit Database Server and Client 64-bit Windows x64



Oracle Database 10g (10.1) Supported Windows Operating Systems


32-bit Database Client 64-bit Windows x64

Oracle9i Release 2 (9.2) Supported Windows Operating Systems


32-bit Database Client 64-bit Windows x64

Oracle 10g - Oracle ODBC Driver is supported only for the Linux x86, Linux Itanium, and Solaris SPARC 64 platforms. (dated July ’06). While I was unable to find anything on Oracles website about ODBC support for x64, I remember this being an issue with two customers and if memory serves they did end up getting a driver that worked from Oracle

Additional 64-Bit ODBC support

“DataDirect Connect64 for ODBC - 64-Bit ODBC Driver for Oracle, DB2, SQL Server, Sybase, Informix - All 64-Bit Platforms”: http://www.datadirect.com/products/odbc64/index.ssp

Documents

In-Memory BI Market