The acceptance of server and storage virtualization has enabled a paradigm shift in how data center infrastructure is purchased and deployed. End user companies are migrating from purchasing separate physical servers connected to storage area networks (SANs) to more modular “reference architectures” which include every component required to run their applications.
Servers and storage were typically sold as separate IT infrastructure elements in the past, usually to different groups within the IT department, and usually under different parts of the IT budget. The move to the cloud has changed all of that. For example, one recent move by a major server and storage vendor enables its storage to directly connect to its blade servers, which may make a storage network unnecessary. You simply purchase the servers, storage and network together as a data center “building block” which converges everything together to make rolling infrastructure a one stop shopping experience.
Everything you need is provided by the modular converged infrastructure, which is now a simple building block for the data center. Traditional storage only vendors have seen the light and are now joining forces with the major network vendors to create the modular building blocks (what I call PODs) which include the storage as a part of the building block.
Since everything is converged into one solution, and virtualized over a converged network, there is no longer a need for a traditional storage only network. The building block (POD) provides storage for both structured (database) and unstructured (file systems) data directly to the servers inside the rack. Even solid state disk (SSD) can be included in the POD as a fast tier of storage for high performance applications. The movement toward modular reference architectures as building blocks for large data centers is now also trickling down to small and medium-size businesses as the vendors shrink their PODs into smaller scale “podlets”.
All of this all got started back in 2009 when converged infrastructure became a driving force to reduce costs, and data center infrastructure management (DCIM) solutions started showing up. For some reason, it seems the large storage companies have become a driving force behind cloud solutions, and many of them see network convergence as a way to beat the other guys into your data center. The storage guys are partnering with the network and server guys to create the data center building blocks of the future.
It remains to be seen whether these developments are ultimately a good thing for the storage guys, as the folks who make both servers and storage may have a leg up in providing standardized solutions with one support perspective. It’s getting scary out there for the SAN folks, as once everything is included in modular racks as part of the POD, then the job of allocating and managing storage may fall to the server or network admins in the future.
I guess it’s always a good idea to keep your skill set and resume up to date. The good news is although reference architectures may make it simpler for end users and cloud providers to create datacenter infrastructure, there will be a surge in the need for people with the more complex skill sets necessary to properly manage and troubleshoot those environments when something breaks.