Storage blows hot and cold. Technologies that seem promising -- even those that manage to rack up some measurable...
market share—can suddenly seem old hat.
Case in point: A lot of industry observers predicted that iSCSI storage systems would shake up the networked storage market and knock Fibre Channel (FC) off its throne, exiling FC arrays to a niche where only the well-heeled shop for expensive, high-performance storage gear. What we've actually seen with iSCSI is more of an evolution of data storage systems: a good idea that, instead of toppling the incumbent, has proven to be more of a gap filler and complement to the high-priced stuff. iSCSI adoption has been slow and steady; according to our surveys, approximately 40% of companies now have at least some iSCSI storage in place. But Fibre Channel still, undeniably, reigns supreme. In our most recent survey, where we asked storage managers about their plans for 2010, of those respondents who expect to buy storage systems, 62% said they'd go for FC while only 22% said they had iSCSI in their sights.
Fibre Channel didn't simply outrank iSCSI on that survey; there were four other storage types separating FC and iSCSI. If iSCSI is going to pose a threat to FC's hegemony, it's going to have to climb over multiprotocol arrays (40%), network-attached storage (NAS) (28%), direct-attached storage (DAS) (26%) and even solid-state storage (23%). So, is iSCSI going away? Nah, but maybe it's not the game changer that so many observers predicted.
And is solid state truly going to outrun iSCSI in 2010 as our survey suggests? I'd guess that in this case the 23% of respondents who said they'll deploy solid-state storage this year represent a lot of pilot projects, a handful of production implementations and -- to steal a phrase from Alan Greenspan -- just a bit of "irrational exuberance" about solid state in general.
Wide-area network (WAN) optimization is another example of the pace of change. It became the storage technology du jour a few years ago but, interestingly, it wasn't specifically intended for data storage environments. It was -- and still is -- primarily an application-enabling technology that just happens to work pretty well when used to back up remote offices by making it possible to send more data faster over relatively thin pipes. But data deduplication now effectively accomplishes the same thing by paring down the data and limiting what needs to hit the network in the first place. As dedupe becomes more ubiquitous, with products that can fit into practically any part of the backup infrastructure, it seems that there will be less need for WAN optimization in backup environments.
And now deduplication is coming out from under the backup covers and showing up in primary data storage systems as well. Although there are only a few players right now, user interest is extremely high given the never-ending disk capacity battle, so 2010 should see more storage vendors getting into the game. Not long ago, thin provisioning was seen as the most promising antidote to bloated disk capacities. Most array vendors, some grudgingly considering the impact it could have on their storage sales, now offer a thin provisioning option. Thin provisioning is a great idea based on the common-sense principle of only giving an application the storage it actually needs. However, the technology hasn't caught on like wildfire; its numbers are up on our storage priorities survey, with nearly twice as many respondents saying they've implemented it vs. a couple of years ago, but the already-deployed figure is still a fairly modest 19%. While 51% say they'll evaluate or implement thin provisioning this year, 60% say they have evaluation/deployment plans for primary storage dedupe. So one has to wonder if thin provisioning will take a backseat as the dedupe juggernaut rolls on.
A lot of these hot-to-cold shifts in data storage are just the effects of the normal churn found in any technology industry. But even if they're just part of the natural selection process, your choices can have significant repercussions down the road. There's nothing wrong with iSCSI, solid-state storage, WAN optimization or thin provisioning -- it's just that none of them is likely to solve all of your data storage problems or revolutionize your shop. Some technologies may seem like the answer to your prayers, but some will never live up to their hype. Finding the best balance of technologies for your environment is tough, but it's what makes managing storage so interesting.
BIO: Rich Castagna (firstname.lastname@example.org) is editorial director of the Storage Media Group.