A colleague recently passed along an interesting article about obstacles converged system+storage (or “Virtual SAN”) vendors such as Nutanix, SimpliVity, and even VMware face convincing customers to migrate from their existing storage systems. The points made are good, but there’s an underlying assumption that migration is acceptable as a primary goal.
It usually is for vendors. No surprise there. But it’s not for most customers.
Having talked with hundreds of CIOs and other senior IT people over the past several years, it’s clear to me that most organizations would be loathe to blindly cram a new storage network design and operational model into their existing infrastructure. It doesn’t matter what the new model is. And Virtual SAN is hardly the first. Another recent example: Private Cloud.
So if migration isn’t an acceptable goal, what is? For more sophisticated organizations, it’s a measured and well-structured pilot. They’d begin by putting less-vital workloads on it. Workloads that can be isolated, and can fail without significant business impact. As the organization gains experience and confidence in the new system, they add more applications.
If the new design and/or operational model continues to work as promised at larger scale, more important workloads can be added. But they would still tend to be new applications. After all, how many IT leaders would deliberately put a vital, currently-operating app at risk? Most wouldn’t if they want to keep their jobs.
That’s why, for most organizations, migration will remain a non-goal for a long time. If the new paradigm ultimately becomes the dominant approach, all new development will go there. As applications are retired, so will their associated infrastructure. Migration will happen if and when the risk becomes so low it’s effectively a “no-brainer.”
This kind of gradual, disciplined approach is what we at Winchester Systems recommend for all but the smallest of IT shops. And the vendors that serve them.