Virtual machine sprawl
You may already be aware of the concept of a virtual machine (VM) – a place on the cloud where you can install software and run applications, and VMs can be configured to do many things, but let’s take that idea one step further. What if you had to manage VMs, all running the same kind of servers and applications, performing similar tasks. Managing those resources – preparing them, destroying them, and then relinquishing them – is another process to consider. You could spend a lot of time configuring the same type of virtual machine over and over again. Inconsistencies might pop up from having multiple “unscripted” deployments. It becomes harder to keep track of VMs, so some might grow “stale” – they are no longer needed, but are still holding on to resources.
IBM Workload Deployer (formerly IBM Cloudburst Appliance) is a purpose-built hardware appliance, designed to answer this kind of problem. Given your cloud infrastructure, IBM Workload Deployer allows you to optimize the image management process for your environment. It dispenses WebSphere Application Server topologies, which are easy to use and customizable, into the cloud.
Patterns and virtual systems
IBM Workload Deployer includes WebSphere Application Server Hypervisor Edition images, preinstalled with operating system, WebSphere Application Server, and HTTP server. These images are ready to use or can be customized with additional software and applications. IBM Workload Deployer also provides patterns, which are templates for topologies that have been optimized for WebSphere environments. They can include multiple virtual images, which are deployed to the cloud as a virtual system. Patterns can be reused to deploy multiple, consistent, WebSphere environments that are middleware-aware. In other words, you can have applications deployed to a cluster in a WebSphere network deployment cell for high availability or failover.
IBM Workload Deployer gives you more than just the ability to spray out virtual machines; it supports the propagation of virtual systems with workload management capabilities into the cloud.
How to get started
You do need to have a cloud infrastructure in place, whether you are using your existing data center or some other hardware or virtualization solution. Configuring IBM Workoad Deployer is a matter of making the appliance aware of that infrastructure:
- Identify the hardware, hypervisors, and networking (IP addresses) for the cloud.
- Choose virtual images from the catalog, and customize them as needed.
- Use patterns from the catalog to describe the WebSphere topology to be deployed.
- Deploy virtual systems in one step.
Learning resources
If you are not already familiar with WebSphere Application Server, there are many courses available to get you up to speed. This page lists several training paths to choose from. Each training path provides recommended courses for a specific product set or job role.
Hi – as a fellow IBMer with some knowledge of this but even more curiosity… can you say something about how this would work end-to-end from a developer's point of view (using dev/test as a use case) – is there an outer shell of governance cycle needed for the virtual image provisioning, monitoring & teardown or is that included in the IWD pattern? And what if the overall system needs non-IBM images? Do we extend the pattern or hand off to another provisioning engine and allow that to orchestrate?
Workload Deployer can indeed dispense generic (non-IBM) images. As for governance, that is a more complex question. It depends on what level you are looking for. It does support the ability to customize images with governance agents.
The predefined patterns cover most typical WebSphere topologies such as a clustered environment, and these patterns can be cloned and customized. The self-paced virtual course that we offer demonstrates how to use and customize patterns – http://www-304.ibm.com/jct03001c/services/learnin…
This article on developerWorks also describes how to customize patterns – http://www.ibm.com/developerworks/websphere/techj…
I can see how this is useful for a data center to meet peak loads. I'm new to Workload
Deployer. Years ago, when virtualization was just coming into data centers, I read that the average server utilitzation rate was something like 15%. So am I right in thinking the main advantage of workload deployer is being able to better utilize hardware resources by dynamically reassigning the hardware to different VMs?
Yes, exactly. Say for example, you have an application that runs in a cluster. You can customize a pattern (that can include multiple images) to reproduce that same workload, and dispense that workload out as many times as needed to scale up, and then destroy the extra workloads later to reclaim and repurpose those resources.
Is the scope of this product to cover all defined images – eg could it cover a mainframe DB2 and PL/I-CICS environment? If that is the case how much reconfiguration is needed for each "clone"? Then the workload on each clone is managed by the product you described?