Upload
hoangdang
View
218
Download
3
Embed Size (px)
Citation preview
Hotel Electronic Network Distribution Association ● 750 National Press Building ● 529 14th Street, NW ● Washington, DC 20045 ● www.HEDNA.org ● Tel. 202-204-8400 ● Fax 202-591-2445
HEDNA White Paper: Push and Pull Connectivity Models for Hotel Distribution
BACKGROUND
The HEDNA Connectivity Working Group (CWG) was formed
following HEDNA‟s 2011 winter conference based on a general
agreement by the represented membership that the challenges of
developing and maintaining multiple distribution interfaces in the
absence of any real standardized implementation is a problem that
the industry needs to address. The group is comprised of members
representing suppliers, distributors and intermediaries that are
committed to achieving the goal of simplifying the way in which they
interact with each other to facilitate the distribution of hotel rates,
availability and reservations through automated interfaces.
INTRODUCTION
The terms “push” and “pull” are applied frequently in reference to the
models of connectivity and data exchange that are employed between
hotel distribution systems. This paper provides an analysis of the
models at both technical and business-logic levels in order to
establish a set of commonly held definitions, provide perspectives on
the way the models are used and offer a comparison of their
similarities and differences.
The HEDNA Connectivity Working Group white paper entitled The
Future of Shopping defined the roles of Data Originator, Data
Publisher and Data Consumer in the communication and
representation of hotel rate and availability data and the development
and maintenance of connections between distribution systems. These
roles apply also in the consideration of push and pull models. This
paper examines how each role is involved in the exchange of
availability, rate and inventory (ARI) information, descriptive content,
and reservation details. It also considers the variety of ways in which
the technical process of data transfer may be established between the
participants in an interface.
The paper includes background information and illustrates the
concepts of push and pull. Its purpose is to serve as a reliable
reference document for hospitality suppliers, intermediaries,
distributors, technology providers, and anyone else with an interest in
the specifics of distribution connectivity.
CONTRIBUTORS
Craig Barnby Vice President of Product Development Pegasus Solutions Tom Bruno Director, eDistribution Development & Operations Marriott International Lew Harasymiw Director of Interface Solutions Sabre Hospitality Solutions Cody Kiker Client Manager/Solutions Engineer DerbySoft Rachel Neal Director of Client Implementations DerbySoft TJ Noble Account Manager, Electronic Distribution Hyatt Hotels and Resorts Dave Revelle Senior Director of Global Partnerships DerbySoft Nader Troudi Director of Technology Expedia Tim Unwin Executive Vice President, Hospitality Solutions RateGain
2
2 Hotel Electronic Network Distribution Association ● 750 National Press Building ● 529 14th Street, NW ● Washington, DC 20045 ● www.HEDNA.org ● Tel. 202-204-8400 ● Fax 202-591-2445
DEFINITIONS
In the context of the hotel distribution methods considered in this paper, the first step is to define the “who”,
“what” and “how” elements which combine to form the complete picture. Roles and methods are outlined
below.
Roles
Data Originator:
Any system in which hotel data is first created or first appears. In practice, a Data Originator is usually
a PMS or a CRS, or possibly a Channel Manager.
Data Publisher:
Any system that accepts data and then re-publishes or redistributes that data. Typical Data Publishers
include CRSs, Channel Managers, Intermediaries, Switches, GDSs, OTAs, etc.
Data Consumer:
Any system that accepts the data and uses it to enable the search for and/or the sale of hotel rooms,
including booking transactions, traffic referrals, data analysis, etc. Examples of Data Consumers are
OTAs, GDSs, search engines, bed banks, etc.
[For additional examples and background see The Future of Shopping]
The roles outlined above are not fixed in the sense that a particular entity may play more than one role
depending on the nature of the transaction and its participants.
Methods
Push:
The Originator or Publisher initiates or triggers the message flow that sends data to the recipient. In a
Push the Originator/Publisher is the active party and the recipient of the data is passive.
Pull:
The Originator or Publisher maintains the current state of the data and the recipient (either Publisher
or Consumer) initiates an inquiry for the data. In a Pull, the recipient is the active party and the
Originator/Publisher is passive.
Hybrid:
A combination of both push and pull elements either a) within a single data exchange or message
flow, or b) over the course of a transaction sequence which achieves a particular business result.
An analogy that can be used to further illustrate the nature of Push and Pull methods is the comparison
between a text message and a phone call. A text message is a push transaction: the source of the message
(or data) is the active party, and the message is delivered to a passive recipient. A phone call, in contrast, is a
pull transaction in that the caller (the recipient of the data) has initiated a request for the data (in this case the
“data” is a conversation) from a passive source.
3
3 Hotel Electronic Network Distribution Association ● 750 National Press Building ● 529 14th Street, NW ● Washington, DC 20045 ● www.HEDNA.org ● Tel. 202-204-8400 ● Fax 202-591-2445
In order to represent the various options and models accurately, consideration of the broad types or
categories of business transactions is also necessary. These categories are defined in detail in the following
section and include: i) transactions related to the maintenance of Availability, Rates and Inventory (ARI), ii)
transactions related to making and confirming bookings, and iii) transactions related to loading, updating and
querying of descriptive content. Note that within a given business transaction or model, each of these
transaction types can be executed by different methods (push or pull) and by different roles (Data Originators,
Publishers and Consumers.)
Transaction types: Availability, Rates and Inventory (ARI)
Non-Cached (Pull and Use): The Data Consumer submits a request to the Data Originator/Publisher‟s system to inquire about
rates and availability for a particular hotel, date, LOS, etc. This response is generated by a consumer
request for use at the time of inquiry. The Data Originator/Publisher responds in turn with the rates
and availability data.
Example: A shopper using a Data Consumer selects a specific hotel, date, and LOS. The request is sent to
the hotel‟s CRS (Data Originator), and an availability response is sent back for the shopper to view and
subsequently make a decision to book or not.
Cached (Push):
The Data Originator or Publisher sends a message containing the change in availability and/or rates
to the pertinent Data Consumer. This updates the data within Data Consumer‟s system so that it
continues to be an accurate reflection of the data within the Originator‟s system.
Example: The hotel CRS (Data Originator) sends an ARI message to the Data Consumer where it is cached.
Any shopping for the hotel will be done on the Data Consumer‟s cache rather than on the hotel‟s system.
Cached (Pull):
The Data Consumer submits a request to the Data Originator/Publisher‟s system to inquire about
rates and availability for a particular hotel, date, LOS, etc. This request is generated by a caching
system, scheduling system, algorithm or shopper‟s search. The Data Originator/Publisher responds
in turn with the rates and availability data that is stored in the Data Consumer‟s cache for further use.
Example: A shopper using a Data Consumer selects a specific hotel, date, and los. The request is sent to
the hotel‟s CRS (Data Originator), and an availability response is sent back for the shopper to view and
subsequently make a decision to book or not. The Data Consumer caches the response for future shoppers
searching the same criteria to avoid redundant transactions being sent to the Data Originator.
Change Discovery Service - CDS (Push / Pull):
Change Discovery is a method of discovering and communicating the existence of change in rates and
availability. Change Discovery allows a Data Publisher or Consumer system to determine whether there has,
or has not been, a change in rates or availability status for a given time frame. With Change Discovery, a
Data Originator, Data Publisher and Data Consumer can communicate with small, efficient messages that
provide a base for building targeted full availability requests for the change detail. This makes each
availability message more useful, meaningful, and efficient, and it effectively eliminates availability requests
for data that has not changed.
4
4 Hotel Electronic Network Distribution Association ● 750 National Press Building ● 529 14th Street, NW ● Washington, DC 20045 ● www.HEDNA.org ● Tel. 202-204-8400 ● Fax 202-591-2445
CDS Push: The Originator or Publisher initiates or triggers the message flow that sends the change notification to the recipient. The Originator/Publisher is the active party as it is responsible for distributing the data, whereas the recipient is passive. CDS Pull: The Originator or Publisher maintains the current state of the data and the recipient (either Publisher or Consumer) initiates an inquiry for the change data. The recipient is the active party in this case, and the Originator/Publisher is passive. More information on CDS can be found in the HEDNA white paper.
Rendered Shopping (Push / Pull):
Rendered shopping allows Originators to provide the rate and availability information in the form of
pre-calculated shopping results, which will be referred to as „rendered shopping data‟, or more
specifically, a „shopping data cube‟. This will contain the rate and availability status information for a
large set of potential room stays, where a room stay is identified by a combination of attributes such
as the property ID, arrival date, length of stay, occupancy, rate plan, room type etc. This means that
shopping requests can be processed by simply using the criteria to look up the rate and availability
status information exactly as it is specified by the supplier in the shopping data cube, without the Data
Consumer having to make any calculations or apply complicated rules. Essentially, this new
approach proposes that the Data Originator communicates explicit shopping data to Data Consumers
whenever a rate or availability status changes, instead of relying on the Data Consumer to calculate
these values based on their interpretation of abbreviated representations of the data.
Rendered Shopping – Push: The data cube is maintained and published by the Data Originator that triggers the message delivery. The Data Consumer is passive.
Rendered Shopping - Pull: The data cube is maintained and published by the Data Originator and the Data Consumer/Publisher triggers the message delivery. The Data Originator is passive.
More information on Rendered Shopping can be found in the HEDNA white paper.
Transaction Types: Reservations
Reservation processing is commonly divided into two categories referred to as Reservation Notification and
Reservation Request. Either push or pull mechanisms may apply to Reservation Notification; Reservation
Request is typically a pull transaction.
Reservation Request (Pull):
In a Reservation Request the Data Consumer/Publisher sends booking details to the Data Originator
and waits for confirmation before confirming back to the guest that the process is complete. It is
literally a request for a reservation.
Reservation Notification (Push):
In a Reservation Notification the Data Consumer/Publisher sends details of a booking that is already
confirmed to the Data Originator. The booking will typically have been confirmed to the guest before
the notification process starts. Literally, the Data Consumer/Publisher is notifying the Data Originator
of a confirmed booking.
5
5 Hotel Electronic Network Distribution Association ● 750 National Press Building ● 529 14th Street, NW ● Washington, DC 20045 ● www.HEDNA.org ● Tel. 202-204-8400 ● Fax 202-591-2445
Reservation Notification (Pull)
In the pull version of Reservation Notification, the Data Originator initiates the process by sending a retrieval
request to the Data Consumer/Publisher. Once again, the booking will typically have been confirmed to the
guest before the notification process starts.
Transaction Types: Content
Static Content:
Static content consists of descriptive information that changes infrequently. This includes property address,
amenities, latitude/longitude, directions, etc. This type of data can be communicated via a push or a pull
transaction set just as with rates and availability data. The frequency of the delivery of such messages is
much less than that of the rate and availability messages because this data changes much less frequently.
Semi-static Content:
Semi-static content consists of data that is necessary to transact an availability message or a
reservation but is not part of the rates or availability. Taxes, guarantee policies, cancellation policies,
rate plan codes, occupancy restrictions, etc. typically comprise semi-static content. This type of data
can be communicated via a push or a pull transaction set just as with rates and availability data. The
comparative frequency of the delivery of such messages corresponds with the frequency of changes
in the data contained within, i.e. semi-static data changes less frequently than rates and availability,
but more frequently than static (descriptive) data.
IMPLEMENTATION MODELS
1. Acquiring ARI Information Using Pull Hotel availability, rates, and inventory may be acquired via any of several Pull models. As noted above, the
primary feature of the pull model is that information is being requested from the Data Originator or Publisher.
Pull (Direct Connect and non-Cached)
This is the simplest model. A Data Consumer requests (i.e. pulls) ARI data by sending availability request
messages to the Data Originator. The Data Originator replies synchronously by sending availability response
messages to the Data Consumer:
6
6 Hotel Electronic Network Distribution Association ● 750 National Press Building ● 529 14th Street, NW ● Washington, DC 20045 ● www.HEDNA.org ● Tel. 202-204-8400 ● Fax 202-591-2445
Pull (Direct Connect and Cached)
This is similar to the previous model except that the Data Consumer caches the ARI data locally after pulling
it. The cache population mechanism may include some form of Change Discovery and/or Rendered
Shopping as described earlier in this document:
Pull (with Publisher and non-Cached)
In this model the Data Consumer does not interact directly with the Data Originator but instead gets the ARI
data from a Data Publisher. The Data Publisher serves as a communication hub in this case, receiving data
from the Data Originator and providing it to the Data Consumer:
7
7 Hotel Electronic Network Distribution Association ● 750 National Press Building ● 529 14th Street, NW ● Washington, DC 20045 ● www.HEDNA.org ● Tel. 202-204-8400 ● Fax 202-591-2445
Pull (with Publisher and Cached)
In this model the Data Consumer does not interact directly with the Data Originator but instead gets the
availability data from a Data Publisher. The Data Publisher serves not only as a communication hub but also
provides a caching mechanism:
2. Distributing ARI Information Using Push
Just as there are several Pull models, any of several Push models may be used for the transfer of ARI data.
In these cases, the Data Originator or Publisher initiates the distribution of information.
Push (Cached)
The Data Originator distributes ARI directly to the Data Consumer. The Originator initiates the flow of data
that is received and cached by the Consumer and used when responding to hotel shopping enquiries. Either
the Originator or Consumer may dictate the format and layout of the data:
8
8 Hotel Electronic Network Distribution Association ● 750 National Press Building ● 529 14th Street, NW ● Washington, DC 20045 ● www.HEDNA.org ● Tel. 202-204-8400 ● Fax 202-591-2445
Push (with Publisher and Cached)
The Data Originator distributes ARI to a Data Publisher who subsequently redistributes it to the Data Con-
sumer. The information flow is initiated by the source in each case. The format and layout of the data may be
dictated by any of the parties in the communication. Note also that the format and layout might be adjusted or
altered by the Publisher mid-flow. Data may be cached by the Consumer or by both the Consumer and the
Publisher, depending on the implementation model:
3. Hybrid Models for Availability
In situations where a Data Publisher is involved in the communication of ARI, the mechanism used by the
Publisher to communicate with the Originator can be different from the mechanism used by the Consumer to
communicate with the Publisher. For example, the Publisher might deploy a cache that is populated using
Pull transactions but then distribute the availability information to Consumers using Push transactions:
9
9 Hotel Electronic Network Distribution Association ● 750 National Press Building ● 529 14th Street, NW ● Washington, DC 20045 ● www.HEDNA.org ● Tel. 202-204-8400 ● Fax 202-591-2445
4. Reservation Requests and Notifications
Reservation processing is commonly divided into two categories referred to as Reservation Request
(generally a pull transaction) and Reservation Notification (for which either push or pull mechanisms may be
employed).
Reservation Request (Pull)
In a Reservation Request, the Data Consumer/Publisher sends booking details to the Data Originator and
waits for confirmation before confirming back to the guest that the process is complete:
Two-phased Reservation Request (Pull)
In some situations there may be two phases to the Reservation Request process:
10
10 Hotel Electronic Network Distribution Association ● 750 National Press Building ● 529 14th Street, NW ● Washington, DC 20045 ● www.HEDNA.org ● Tel. 202-204-8400 ● Fax 202-591-2445
Reservation Notification (Push)
In a Push Reservation Notification, the Data Consumer/Publisher initiates the process by sending details of a
booking that is already confirmed to the Data Originator. The booking will typically have been confirmed to
the guest before the notification process starts:
Reservation Notification (Pull)
In the pull version of Reservation Notification, the Data Originator initiates the process by sending a retrieval
request to the Data Consumer/Publisher. Once again, the booking will typically have been confirmed to the
guest before the notification process starts:
11
11 Hotel Electronic Network Distribution Association ● 750 National Press Building ● 529 14th Street, NW ● Washington, DC 20045 ● www.HEDNA.org ● Tel. 202-204-8400 ● Fax 202-591-2445
USE CASES AND ADDITIONAL CONSIDERATIONS
Data Consumers generally fall into one of two main categories, depending on their mode of operation:
“Seamless” processing, as employed by GDSs and some OTAs, in which pull transactions predominate;
Extranet and cache configurations, used by the majority of web-based Data Consumers, in which ARI
updates and reservations are typically processed using a mixture of push and pull transactions.
Some Data Consumers may be in both categories at different times. One of the key differences between the
categories is that, in the case of ARI and descriptive content, seamless processing tends to involve a
centralized or “single source” view of information, whereas the use of extranets and caches leads to the
replication of information in multiple locations.
For Data Originators, especially those with large portfolios of hotels and substantial volumes of data to
maintain, the use of the pull model for ARI and descriptive content is considered preferable because it
generally allows more granular control over factors such as last room availability or the application of yield
management restrictions. However, because the configuration of many Data Publishers and Data Consumers
is more compatible with push transactions, the decision for Originators between Push and Pull is not simple.
Channel management products evolved as a method of automating the maintenance of information on
multiple Data Consumer extranets. The first channel managers were built to push ARI to a number of
destination sites, each of which had its own representation of rates and room types to be sold. As a result,
the primary function of early channel managers was to provide the mapping between the Data Originator‟s
data structure and that of the Data Consumer.
Over time, the role of channel managers has evolved in sophistication to include, for example, the delivery of
reservation notifications from Data Consumer to Data Originator. This in turn has led to more elaborate
deployment models and allowed channel managers to offer another layer of control in the distribution
process.
In today‟s environment, the range of Data Publishers has expanded to include a variety of intermediary
technologies that are able to extract and deliver data to and from Originator and Consumer systems (using
both push and pull transactions). These technologies can also store, manipulate, and translate the data
according to the business processes being performed, and then redistribute the data in a format appropriate
to the recipient. These functions of collection, translation and delivery facilitate the communication between
Originators and Consumers and reduce the need for one party to adapt technology and processes to match
the other.
The entire process of managing ARI, content, and reservations has become increasingly complex and
inefficient for Data Originators, Data Publishers, and Data Consumers. The volume of shopping requests
continues to grow at a rate that would be unmanageable for most of the systems that originate the data to
answer directly without the buffer provided by the databases and caches of the systems that consume it. At
the same time, the volatility of the data is also increasing, which magnifies for Data Originators the task of
keeping all of the data they distribute current.
The challenge presented by increasing volumes of both shopping requests and changes in ARI data is further
complicated by the lack of adoption of a standardized representation of rates and availability. This results in
12
12 Hotel Electronic Network Distribution Association ● 750 National Press Building ● 529 14th Street, NW ● Washington, DC 20045 ● www.HEDNA.org ● Tel. 202-204-8400 ● Fax 202-591-2445
both Data Originators and Data Consumers needing to manage these volumes across multiple different
message formats. Yet another complication is that not all Data Consumers can easily support or replicate the
variety of restrictions and business rules required by increasingly sophisticated Data Originators.
This lack of consistency in functionality and business logic across various systems leads to price and
availability disparity as products that appear on one site may be priced differently on another or may be
missing altogether. These disparities disadvantage both the Data Consumers and Data Originators, and they
can be exacerbated if a Data Publisher is involved as this adds an additional layer of complexity and another
party‟s system, the potential limitations of which may negatively impact the distribution flow.
For more information about HEDNA, visit HEDNA.org, email [email protected] or call 202-204-8400.