Cloud image consolidation
Antonio_Di_Cocco 060001977Q Visits (1959)
Using IBM SmartCloud Provisioning end users can easily create and use new virtual machines without taking care about how they are running and where they came from. Users just pick up an image to be deployed from a catalog and run it. These means that in order to fulfill all requests the catalog should be as wide as possible. Theoretically it should contains all possible combination of Operative System plus software. Meaning that cloud administrators must manage thousands and thousands of image. Number of base images increase quickly improving management costs, which may lead to a much more expensive solution rather than a cost saver infrastructure. Looking to INTERNET, there are several suggestion to consolidate your image catalog to be as smaller as possible.
Best way to create and to maintain a small image catalog is to create few standard configuration depending on the user's job role. For example a developer could require an Ubuntu systems with Lotus Notes and Rational Software Architect installed on it, while a tester may need a Windows system with TEM agent, DB2 and some other middleware to run his test scenario. In this case we can define a standard so that any user can ask for end usage driven Virtual Image deployment. She/he will require to deploy an image not selecting it by its content, but based on its job.
Even if it is a good suggestion, it could not be easy to implement. In fact if you are creating your cloud solution from scratch you can force end user to select in a small catalog the best image fitting their requirement. But if you already have your cloud environment up a running for a while and your image catalog is already out of control, its consolidation could not be an easy job. Cloud administrators should open all virtual images to look into them, understanding their content. And than decide which is the most representative, making them master templates. Just think to do this job for thousands of image.
Luckily IBM Virtual Image Library will help you in this work.
One of key features for IBM Virtual Image Library is the capacity to introspect virtual image understanding their content and allowing loud administrators to compare between them. In this way they will be able to understand how much two images are similar. When you register an operation repository to IBM Virtual Image Library, a discovery process starts. During this phase information about all images and virtual machines are retrieved from remote repository. At this time only meta data about remote objects are stored locally. Moreover virtual images are indexed, reading their contents remotely.
Once registration has finished you have possibility to start working with your catalog. Most useful operation are:
As you can imagine having a so powerful tool it will be an easy job to define your master images (defining standard configurations). And to consolidate them by merging similar images reducing them to a smaller and manageable set.
Core function for previous feature is IBM Virtual Image Library capability to introspect remote images. There are two types of analysis available:
In both cases IBM Virtual Image Library will not copy any image locally, but simply connect to remote hypervisor data store to read image disk. Reducing time and network traffic spent during this operation.
IBM Virtual Image Library allows you to introspect remote images to consolidate their number removing the unnecessary one. One you consolidated them you can also import the smallest catalog into IBM Virtual Image Library reference repository allowing you to move across hypervisors. As mentioned in Image portability across hypervisors