How does using hyperconverged infrastructure change my hardware refresh cycle?
[dropcaps type=’square’ color=’#ffffff’ background_color=’#e04646′ border_color=”]O[/dropcaps]ne of the more painful aspects of enterprise IT is that thanks to Moore’s Law, data center administrators find themselves replacing hardware every couple of years. Depending on the current configuration of the data center, this can be a major time drain at best, and a massive headache at worst (including outages, late nights, and even data loss).
Thanks to virtualization abstracting running workloads from the underlying compute resources, and with the support of technologies like VMware’s EVC, vMotion, and Storage vMotion, a hardware refresh can be less painful than it was a decade ago. However, major refreshes are still plagued by potential interoperability issues, opportunity for human error, and limitations in platform design. As an example, swapping out storage controllers for bigger ones on a monolithic storage array tends to be a real chore.
Another challenge – especially at the executive level – is budgeting for which hardware refreshes will come due at a given time and making sure that all the disparate resources get uplifted at the proper time. The storage array needs refreshed in Q2 of this year, followed by the deduplicating backup appliance in Q3, compute hardware in Q4, and so on. Budgeting both money and time can become quite a cumbersome process, and all of this is simply to stay afloat. This isn’t even progress.
Hyperconvergence can have a dramatic impact here, as the converged architecture makes upgrading a snap. Because the SDS component of an HCI solution fully abstracts underlying storage hardware from the storage platform presented to the hypervisor, refreshing storage is easy now too. Rather than keeping 20 different plates spinning in a legacy architecture, wouldn’t it be nice to purchase a pile of HCI nodes and a couple of switches and be done with it? The burden on both CIO/CTO’s and the IT administrators doing the work is significantly reduced when an organization is fully shifted to a hyperconverged model.