Heterogeneous Foundations for Cloud: Simply Overrated

I spend a lot of time discussing VMware’s vCloud solution stack to many different customers, each with varying objectives when it comes to their cloud journey. The majority of them fall under two groups – Group A) those who know what they want and where to get it and Group B) those who think they know what they want and have been shopping for the “right” solution since before cloud hit the mainstream – one “cloud bake-off” after another while changing requirements in real-time. Can you guess which ones meet their objectives first? Hint: it’s the same group that delivers IaaS to their enterprise and/or customers using proven technologies and trusted relationships in the time it takes the other to host a bake-off.

For group A the requirements are straightforward – deliver me a solution (and technology) that meets exceeds all the characteristics of cloud [see: defining the cloud] so I can transform my infrastructure and deliver next generation IT to the business. Sound familiar? It should because this is where the greater majority is – whether they accept it with open arms or are trying to meet agency mandates (or both). These are the organizations that understand the value of a COTS solution that promises to reduce cost, complexity, and time to market. These are the folks that consider what has worked so incredibly well in the past and stick with it. They look at the foundation that has built their virtualized infrastructures and helped them achieve unprecedented levels of efficiency, availability, and manageability. These are vSphere customers (did you see that coming?). Remember what the very first characteristic (and prerequisite) of Cloud is – Pooling of Resources. More than 80% of the virtualized world is running vSphere as their hypervisor of choice. In fact, a new VM is powered up on vSphere every 6 seconds and there are more VM’s in (v)Motion than there are planes in the sky at any moment. There is no question that VMware’s flagship hypervisor has changed the way we do IT – a hypervisor that has earned the right and reputation to be your cloud’s foundation…vCloud’s foundation.

But not everyone gets it (enter group B). These are the folks that set requirements they think are intended to benefit the business or customer and end up burning resources, money, and time in the process. My job is to look at the business’ objectives, understand their unique requirements, propose a solution, and help determine the resulting architecture. But every once in a while a customer throws out a requirement that just doesn’t make sense…and I feel, as trusted advisor, it is my responsibility to make sure they understand the impact of such requirements.

This brings me to the topic of this post and the most often misguided “requirement” out there: “my cloud needs to support heterogeneous hypervisors”.

Say what!? Heterogeneous hypervisors? I’ll just put this out there – VMware’s cloud framework (specifically vCloud Director) does not support heterogeneous hypervisors – and for a very good reason! What benefit will this provide when there’s an opportunity to build this baby from the ground-up? Let me be clear about one thing – the need to support a heterogeneous anything is a problem and not an effective business strategy. Heterogeneity often occurs when IT merges – whether that’s in a datacenter consolidation, business merger, bankrupt vendor, whatever. The business typically wants to save existing investments and needs a new way to manage those assets in a centralized/consolidated manner. A great example of this exists in the storage world – as datacenters were consolidated and several different flavors of storage subsystems were expected to play together, storage virtualization solutions were needed to make it so. There are several solutions out there to choose from – IBM SAN Volume Controller (SVC) or NetApp V-Series just to name a couple. Bottom line is the organization gained a heterogeneous storage environment and needed a solution to bring it all together and achieve centralized management. Although there are solutions available to help (some better than others), they are really just a band-aid and still result in everything you’d expect from such a situation:

These are all results of a problem. Organizations rarely choose to add complexity, cost, risk, etc. to their infrastructures but instead employ available technologies to help reduce the pain of such a situation. However, as the environment scales, these same organizations do choose to scale the native capacity first in an effort to avoid making the problem worse (ex: adding a storage capacity that natively integrates with the front-end solution).

When it comes to building a cloud infrastructure, most organizations are early in the design and planning process and have an opportunity to employ proven technologies and gain seamless integration, high efficiencies, centralized management, etc. all on top of a solid foundation. The key word here is foundation (i.e. the hypervisor) – the most critical component in this architecture. Why would any organization choose to take the heterogeneous approach and deal with the added risks when so much is at stake?

And finally, for all you out there that suggest that not supporting a heterogeneous foundation creates cloud vendor lock-in (that’s you, group B), I only have this to say: regardless of who you trust to be your hypervisor, your best bet is to select a solution that provides an open extensible framework, exposes API’s for seamless infrastructure integration, and has the trust and reputation your business or customer needs. I won’t name names…but there’s only one.