|By Joe Arnold||
|February 24, 2014 08:45 AM EST||
"Software-defined," like any new trend that technology companies rush to attach to has suffered from marketing hype. Starting in mid-2012 with the acquisition of Nicera by VMware, most traditional infrastructure technology vendors across compute, networking, and storage have some messaging around how software-defined fits into their product strategy.
But software-defined is not a traditional concept. Since the transition from mainframe to distributed computing, corresponding with the rise in networking, most technologies in the data center have been very hardware specific. For many years, the way to get the right amount of intelligence in the right place to execute the functionality has been with specialized hardware.
Software-defined is fairly self-explanatory. The value is in the software, with one of the biggest benefits being the use of standard hardware that is a fraction of the cost of vendor-specific hardware, which has been the norm for data centers since the beginning of the 21st century. Standard hardware from servers to networking devices have so many resources available that specialized hardware no longer provides the differentiation it once did.
Freedom from hardware opens up freedom to extend SDDC outside the walls of the data center. SDDC helps organizations deliver a modern private cloud in the same model that large operators like Amazon and Rackspace use to deliver public cloud. What's more, using the right software-defined technologies enables hybrid cloud, the panacea for most enterprises. This is one of the main reasons that open source technologies, and those with very strong standards are emerging in the software-defined space.
That said, software defined has the biggest opportunity to benefit private data centers, where the majority of applications simply cannot run in the public cloud based on policy or preference. Here are some best practices when looking at how software defined can benefit your private data center architecture:
Take advantage of your already efficient procurement: Odds are your organization has a go-to vendor or reseller of servers. Whether from HP, Dell, or a white box vendor like Supermicro, the premium paid on server infrastructure is much smaller than with storage. You may even have a volume purchase agreement, making purchasing of hardware for software-defined storage even more affordable. Buying infrastructure for SDDC is no different than your channels for standard servers and networking equipment.
Insist on standards compliance to avoid future lock-in: SDDC is a real opportunity to reduce or eliminate lock-in among the technologies used in your data center. The best software-defined solutions are based on industry standards so freedom to change is retained. This is more than basic interoperability with a standard API, as that case still relies on the software vendor to keep up with changes. Avoid software-defined solutions that are vendor specific and limit your flexibility to integrate and innovate as this market continues to evolve.
Have a preference for open source technologies: Once dismissed by their proprietary competitors as immature, open source operating systems, middleware, application frameworks, and databases are now standards in the data center. The same trend will hold true for software-defined solutions. Using an open source technology does not preclude organizations from working with commercial vendors to support the success of open source in the data center architecture.
Where are the biggest opportunities to use software defined? We believe it is in storage, a solution area where very large premiums have been paid for many years, based on the perception that specialized hardware was the only way to keep data safe and available. The biggest operators have proven the opposite - they can be available to millions of concurrent users without downtime or losing data while using standard hardware and intelligent software.
The reality is data is simply growing too fast and must be retained too long, at a cost that in many cases must be as close to zero as possible. Plus, unstructured data is the fastest growing, fueled by SaaS applications and the transition to mobile devices. Any private data center today is in direct competition with the operators of large public clouds. Internal users demand the flexibility and operating costs the big operators have proven are possible, so private operators must use the same software-defined strategy to remain competitive.
"Utility Computing" was a hyped trend at the start of this century that was before its time. Some might classify SDDC in the same category, all hype. Objectively, SDDC is a natural extension of cloud, and should prove to be equally as disruptive to the data center architecture as cloud has been. Software used for compute, networking, and storage will need to be as standard as the servers you buy today in order to fit into your data center architecture of the future.