This article can also be found in the Premium Editorial Download "Storage magazine: Everything you needed to know about object-based storage."
Download it now to read this article plus other related content.
Moving forward, I said that a bit of consideration at the time of infrastructure planning regarding how data would be protected could make all the difference between a smooth recovery or lots of delays and disappointments and, well, disasters. Hardware lock-ins, which are created when relying on proprietary on-hardware snapshot, mirroring and replication technologies purchased with the storage rigs themselves, are problematic over time. For the functionality to work at all, you usually need to buy two copies of the rig from the vendor: one for the primary site and another for your recovery facility. These specialized pairings get in the way of coherent data protection and recovery, especially as data storage infrastructure grows more heterogeneous.
The good news, I offered, is that software-based storage virtualization can help to alleviate this problem by enabling delivery of protection services such as continuous data protection (CDP), snapshot, mirroring and replication services on a cross-platform basis -- irrespective of hardware brand. Preventing hardware lock-ins from the outset is key to making disaster recovery efficient. A good storage virtualization approach can provide a more unified way to deliver the right data protection services to the right data at the lowest possible cost.
Some confused looks came my way as the attendees slathered butter and salty sauces onto their potatoes and vegetable. After all, they spent big dollars on storage
These concerned expressions deepened as I offered another quick caveat: Not all storage virtualization software is the same. When VMware adds a storage hypervisor microkernel to its stack later this year, in all likelihood it won’t deliver the finely tuned service resiliency and platform agnosticism of more tenured storage hypervisors already in the market. Instead, for the VMware stuff to work, you would need to have all servers running the VMware hypervisor, which is essentially the same dog you get from buying single vendor lock-in hardware, but with a different set of fleas.
Citing examples, I explained that the complexity of recovery is made worse by the proliferation of data protection service lock-ins. If you think it through from the start -- when building infrastructure itself -- you can create a rich, built-in, hardware-agnostic, data protection-enabled infrastructure that will cost a lot less than bolting on various mirrors, replicators, clusters and backup processes after the fact.
My time was up, aperitifs and gooey chocolate cake was served, and the sponsor took the stage Yes, the right way to build disaster-proof infrastructure might be the approach that Toigo spelled out (the sponsor noted), but for the rest of us, the affinity for a particular vendor’s gear, pre-integrated with the functionality you need, is often simply the more preferable way to buy storage gear. Rather like buying a prepared dinner at a nice restaurant instead of cooking a healthy meal yourself featuring judicious limits on fat, starch, sugar, salt and other ingredients.
We opt to exercise afterwards rather than designing a healthier diet up front. So it goes.
This was first published in June 2012