2016 - The Year of the Storage Wars

The Geeksultant's picture

There are so many players in the storage and the hyper converged market segment that it's almost impossible to choose one. At the end of the day, for most customers, it still comes down to cost. "What am I going to pay for this?" Storage has always been one of the largest expenditures of any data center. Still true today as it was decades ago. Today's storage is faster and slimmer and has more tricks up it's sleeve with deduplication, high availability and automated backup built in. On top of that, many competitors have gone the software route utilizing SD-Storage (Software Defined) such as VMware's V-SAN product that allows you to use local storage in your servers as a grid style solution served up as an aggregate to all servers in that cluster. Other players, such as Atlantis and it's ILIO product, which utilizes server RAM before the storage I/O ever hits actual disk or flash based storage, have also reduced the need for standard or even hyper converged compute and storage offerings. Citrix's PVS with "Cache in RAM with fallback to disk" has also reduced the need for high end, and for that matter, high cost back-end storage for VDI workloads such as XenApp and XenDesktop. A few years ago, the idea of hyper converged products gained ground quickly due to the data center's need for reducing it's space, electric and heat footprints. It started with blade centers and blade servers. The idea was that one can put 8, 16 or more blade servers in the same space of what used to be taken up by three or four rack servers. Today, I see many customers moving away from blade centers and blade servers and gravitating back to individual rack servers. One, they use less electricity and produce less heat than prior generations and many customers now feel that blade centers have themselves become single points of failure. However, I think the reason that this trend continues, is that blades are limited on what local storage they can carry. Traditional rack servers can now be loaded with cheap 15k disks or even SSD drives whose prices have drastically fallen in the past 12 months. Now add in the new software defined storage solutions, and customers can build their own hyper converged solutions for far less than any of the fully integrated purchased options. It's the same phenomenon that occurred when the first PC's came out in the 1980's. It didn't take long for end users to figure out how to build their own clone PC for less money and often with better performance than what the major manufacturers had to offer. Storage needs will continue to drive innovation and manufacturers, with their R&D budgets, will always lead in what is the next and greatest thing in storage or hyper convergence. Then, as always, someone will give up the magic sauce and the customers will build their own "clone" for less cost and then the cycle will continue. Technology will continue to evolve, and customers will continue to find the cheapest way to duplicate it.