Pods fit for virtualization, but bring HA clustering challenges

Converged infrastructure pods are a good option for virtualization, but administrators must decide whether simplicity is worth a clustering headache.

Converged infrastructure pods offer numerous benefits, but they may not be the best fit for every virtual infr

astructure.

The basic idea of a pod is that server, storage, and networking components are bundled together and sold a single integrated product.

Obviously, the big selling point for pods is supportability. The components included in a pod have been tested to ensure they work together. Since the components are sold collectively as a pod, getting hardware-level support should be relatively easy, but the benefits associated with using pods go beyond supportability.

Another big advantage to pods is consistency. Because pods adhere to a predetermined design, they are completely predictable. There shouldn't be any surprises in the deployment process and the pod should offer a predictable level of performance.

Another advantage to using pods is that some (but not all) pod vendors work directly with hypervisor vendors, such as VMware or Microsoft, to make sure that the pods are optimally configured for use as a server virtualization platform. Furthermore, the components within a pod are usually performance matched to work well together and to avoid major hardware bottlenecks.

Pods are not one size fits all
Of course critics are quick to point out that pods use a rigid design that is largely inflexible. As a result, an organization might end up paying for resources within a pod that they really don’t need.

The rigidity of a pod varies by vendor. Some vendors offer relatively few pod customization options, while others treat pods as a modular architecture and allow customers to pick and choose the components that they want to include in the pod.

Another common criticism for pods is that using pods leads to vendor lock. On the surface, this argument doesn’t hold up. After all, just because an administrator has purchased a pod from a vendor does not mean that they can’t return to a traditional rackmount design and use a different vendor the next time they need compute resources.

Even so, it is important to remember that one of the big selling points behind pods is that most pods include management software that is specifically designed for the pod. If an organization mixes and matches hardware from various vendors then they complicate the management process.

When it comes to using pods for server virtualization, there are advantages and disadvantages. One advantage is that it is easy to build a pod into a self-contained host cluster. The cluster can make use of local storage contained within the pod or it can connect to a centralized storage device. As the organization grows, they can simply deploy more pods, configuring each pod as a separate host cluster.

One benefit to this approach (besides simplicity) is that many pod vendors offer different options for pods. For example, a vendor might offer a “virtualization pod” and a “high performance pod.” Different types of pods can be used for different purposes. An administrator might use a high performance pod to run virtualized database servers, while a more modest pod handles less resource-intensive workloads.

The modular nature of pods can certainly make it easier to scale the organization’s virtualization infrastructure than it otherwise would be. Even so, there is at least one big disadvantage to using pods for virtualization.

HA Clustering considerations for pods
Unless a pod offers full redundancy for all components (including power supplies and cooling), the pod can become a single point of failure. If a pod were to fail, then all of the servers within the pod would fail. If the pod were being used as a self-contained host cluster, then the virtual machines would not be able to failover to other hardware because all of the hardware in the contained cluster is running in the pod. This isn’t to say that you should avoid using pods, but rather that you should make sure to design your host cluster in a way that prevents this type of situation from happening.

Like any other IT technology, pods have their good points and their bad points. Even so, pods are often a great option for running virtualization hosts, especially in organizations that are looking to reduce complexity.

One of the major benefits to pods is that they are modular. Administrators can pick and choose the components that they wish to include in the pod in order to best achieve the pod’s intended purpose. While the argument could be made that the same thing is possible with a standard rackmount system, pod components are certified to be compatible with one another and are usually performance matched, so as to reduce hardware bottlenecks.

This was first published in April 2014

Dig deeper on Virtualized clusters and high-performance computing

Pro+

Features

Enjoy the benefits of Pro+ membership, learn more and join.

Related Discussions

Brien Posey asks:

Do you think the advantages outweigh the challenges of using integrated infrastructure pods for virtualization?

0  Responses So Far

Join the Discussion

0 comments

Oldest 

Forgot Password?

No problem! Submit your e-mail address below. We'll send you an email containing your password.

Your password has been sent to:

SearchVMware

SearchWindowsServer

SearchCloudComputing

SearchVirtualDesktop

SearchDataCenter

Close