Debunking the Top Data Center Myths
For all the changes taking place in storage these days – solid state, virtualization and even software-defined storage networking – it is easy to overlook the fact that these advances are placing the development of storage platforms on a very different trajectory than in years gone by.
Within traditional infrastructure, capacity was the name of the game for storage. Larger volumes and bigger loads meant the enterprise was in a constant race to ensure they had a home for every bit. And that meant hardware, and lots of it.
Today, speed and agility are top priorities. With cloud resources available to take up the slack when necessary, the need to push bits in and out of storage in ways that suit advanced virtual architectures is paramount. And the development arcs of many leading solutions reflect that fact.
PernixData, for example, recently released a new version of the FVP solution designed to more closely align storage resources with the traffic patterns and I/O characteristics of virtual applications and workloads. The system works by introducing pooling and clustering capabilities to available RAM and Flash resources, providing not only accelerated storage service to VMware ESXi environments but broad scalability as well. The new version adds support for both file and block protocols and adds synchronous and asynchronous replication for improved fault tolerance and recovery capabilities.
Meanwhile, HP has added new Flash capabilities to its XP7 disk array in a bid to bolster the performance, scalability and resilience of advanced virtual architectures. The system is now said to provide six-nines reliability and scales to 4.5 PB using a mix of Flash, SSDs and SAS hard disks. Most importantly for advanced virtual environments: Top throughput has been clocked at 3 million IOPs with latencies of less than a millisecond.
For many enterprise architects, however, the question is not where storage is headed, but what will happen to existing storage infrastructure. According to Wikibon, traditional SAN and NAS infrastructure will go the way of tape; that is, still in the mix but no long the heart and soul of the storage environment. The company’s own research suggests a steadily declining SAN/NAS/DAS market over the next decade or so, with server-based SAN and modular hyperscale technologies poised for dramatic upswings. The thinking behind this is that the SAN and NFS-based systems simply cannot provide adequate service to today’s virtualized, multicore, multi-socket server platforms.
Naturally, this has some people questioning how long it will take for these upheavals to play out on the bottom lines of today’s leading storage vendors. Forbes’ Peter Cohan, for example, notes the steadily eroding share value of companies like EMC even as the amount of corporate data, and thus the need for storage, has been mounting steadily. Could it be that EMC’s continued dominance of the external drive market merely means it remains tied to yesterday’s storage environment, and that even moves like the Emerging Storage unit and the Pivotal analytics software endeavor merely place the company as one of many vying for the new storage sector?
EMC executives will undoubtedly argue that point, but it can’t be denied that in times of great change, size is not always a predictor of success. Just ask Bear Stearns.
Storage consumers have their own rows to hoe when it comes to transitioning to the future. And if that future involves nimble, dynamic data functionality, there is no room in that world for storage architectures that cannot match the performance and flexibility emerging in the server rack or on the network.
