The terms “software-defined data center” (SDDC) and “software-defined storage” (SDS) are commonly thrown around, typically being associated with the abstraction of core infrastructure functionality into a common software plane that can then be deployed on low-cost, commodity hardware. This definition misses an aspect that is critical to achieving the true value of the software-defined data center – automation for programmability.
Modern businesses require extreme levels of reactiveness to employee and customer requirements, to maximize productivity and differentiation. IT infrastructure must be as streamlined as possible, to move at the speed of agile DevOps teams that are programming and collaborating in real time to update and bring to market new applications in response to these business needs. Against this backdrop, cumbersome IT management tasks – for instance, requiring 15 or more steps to provision an individual machine – are a core bottleneck to business productivity.
While automation has become more commonly baked into the compute and networking elements of the SDDC stack, most SDS platforms today have only taken the first step of virtualizing controller functionality. They are effectively a legacy, hardware-driven architecture that has been ported over to the controller, as a result lacking automation.
A true SDDC storage solution should provide not only common management across storage media, protocols and deployment models, but also integrate API-driven programmability to enable automated resource substantiation and orchestration across these infrastructure resources. Self-service provisioning, policy-based optimization and tiering, as well as the ability to more quickly integrate new infrastructure resources are characteristics to look for.
Looking ahead, artificial intelligence (AI) and machine learning (ML) will effectively facilitate the next stage of infrastructure reactiveness and optimization. Workload requirements are becoming more diverse, more dynamic and more data-driven. SDDC storage platforms that integrate AI and ML will be better positioned to respond by applying algorithms to identify and predict workload patterns. The result is more finely-tuned and more independently-tuned storage resources.
Cumbersome management tasks not only inhibit business reactiveness, but they also add substantial cost to the storage infrastructure that, over time, substantially exceed the upfront purchase price of the storage hardware itself – running contrary to the core value proposition of the SDDC. Access Storage Switzerland’s and Datera’s webinar, Will the Software Defined Data Center Ever Happen?, for more on how to overcome storage as the roadblock to a true SDDC implementation.