Upload
diana-parks
View
215
Download
0
Tags:
Embed Size (px)
Citation preview
Two Key Enterprise Technologiesfor the Cloud
Message Oriented Middleware (MOM)
Allows load sharing and loose coupling and is robust to server failure.
Virtualization:
The ability to run multiple operating systems on a single physical system and share the underlying hardware resources
G53ELC
A definition
Cloud Computing:
“The provisioning of services in a timely (near instant), on-demand manner, to allow the scaling up and down of resources”
Requirements…
IApplications are International and expect users from anywhere in the world
Applications access huge (Petabyte) databases
Applications expect to download content rapidly anywhere
Applications need to scale with user load without degrading response
Applications need to be available 24/7 365 days a year
Applications need to be secure and well defended
Companies wish to pay only for bandwidth and server time used
What’s a Petabyte?
1024 Bytes = 1 Kilobyte
1024 Kilobytes = 1 Megabyte
1024 Megabyte = 1 GigaByte
1024 Gibabytes = 1 Terabyte
1024 Terabyte = 1 Petabyte
Exercise for the student: Write down the number of bytes in a Petabyte as a number
Petrabyte applications are not unusual nowadays
Google believed to processes 30 PB a day
eBay has 7 PB of user data
Facebook has 36 PB of user data
Suppose you are Forbes.com
You offer on-line real time stock market data
Why pay for capacity weekends, overnight?
9 AM - 5 PM,M-F
ALL OTHER TIMES
Rate of Server
Accesses
Forbes' Solution
Host the web site in Amazon's EC2 Elastic Compute Cloud
Provision new servers every day, and deprovision them every night
Pay just $0.10* per server per hour * more for higher capacity servers
Let Amazon worry about the hardware, the scaling, the local (edge) delivery, the security, the availability, and the backup(?).
Cloud computing takes virtualization to the next step
You don’t have to own the hardware
You “rent” it as needed from a cloud
There are public cloudse.g. Amazon EC2, and now many others
(Microsoft Azure, IBM, Sun, and others ...)
A company can create a private cloudWith more control over security, etc.
G53ELC
Lower Cost
No need to pay for infrastructure up front
No need for expensive support staff
only pay for what you use
Great for start-ups – may even be free
G53ELC
More Agile
It used to take 3 months to set up an application on a cluster of servers
Takes half an hour in the cloud
Scale up or down (elasticity)
OK It’s a good idea. How does it work?
We already saw how to set up a RESTful web service on the Amazon cloud. It took five minutes.
Very easy to serve static web pages in this way
Quite simple to store and access data in the cloud as files or databases
More tricky to set up large scalable applications, but this is where really big pay-offs are possible.
G53ELC
How Cloud Computing Works
Various providers let you create virtual servers Set up an account, perhaps just with a credit card
You create virtual servers ("virtualization") Choose the OS and software each "instance" will have It will run on a large server farm located somewhere You can instantiate more on a few minutes' notice You can shut down instances in a minute or so
They send you a bill for the processor time and comms bandwidth that you use
G53ELC
Worries?
How do I pick a provider?
Is my data secure?
Do I have any control over where my data is moved to?
How can I be sure the provider will live up to all those promises?
(footnote)How come Amazon?
It arose out of efforts to manage Amazon’s own services (Each time you get a page from Amazon, over a hundred servers
are involved) See reference Amazon Architecture on ELC web page
They got so good at it that they launched Amazon Web Services (AWS) as a product
G53ELC
Cloud Computing Status
Seems to be rapidly becoming a mainstream practice
Numerous providers Amazon EC2 imitators ... Just about every major industry name
• IBM, Sun, Microsoft, ...
Major buzz at industry meetings
G53ELC
The revolution
Rent it instead of build it – pay for what you use
Rely on the experts to solve all those worries..
There is a major revolution underway in how we manage hardware Use many servers with virtualization Applications organized with MOM
Data cached close to delivery point
Deployment and monitoring are in-house functions