USING SOFTWARE ABSTRACTION AND COMMODITY STORAGE HARDWARE IN HYPER-SCALE CENTERS
By Farid Yavari
In our previous blog post , we covered how to build a disaggregated storage model out of commodity hardware for lowest Total Cost of Ownership (TCO) in today's hyper-scale data centers. Deploying commodity hardware usually requires an intelligent orchestration and management software like FreeStor ® to provide an abstraction layer separating heterogeneous storage hardware from the applications.
FreeStor enables deployment of storage technologies with different capabilities, at various cost points, in a tiered model. With FreeStor, data is placed on the right storage media, which provides the Service Level Agreement (SLA) -driven services to apps and achieve the lowest possible TCO. Combining heterogeneous commodity storage with an intelligent policy-driven software-defined storage (SDS) also provides flexibility in data migration, enables seamless tech refresh cycles, and independent scaling of the storage and server infrastructure. As depicted in the diagram below, having a comprehensive Representational State Transfer (REST) Application Programming Interface (API) interface to interact programmatically with the Intelligent Abstraction® layer, provides an industry standard approach to managing the SDS environment, while enabling automation in deployment, monitoring, and management of the storage infrastructure.
Some data protection and high availability (HA) capabilities aren't necessary for a scale-out storage architecture because often applications that are running in these environments feature built-in resiliency and self-healing capabilities. Other features such as application-aware snapshots, clones, deduplication, and compression are tremendously valuable depending on the types of workloads.
Private and public cloud capabilities, and having the right drivers to interface seamlessly with the cloud, are an essential feature to look for when evaluating FreeStor or any other software-defined storage solution. Being able to place data in the cloud, and access it as needed, is a requirement in most hyper-scale data centers. Complete storage abstraction software needs to operate in a virtualized environment as well as bare-metal implementations supporting block, file, and object protocols.
There are many challenges involved in architecting a resilient, scalable storage infrastructure for the next generation of hyper-scale modern data centers. Storage disaggregation for scaling and tiering purposes needs to address density, cost, and performance for various use cases. FreeStor helps companies get the most out of these disaggregated storage models, and meet the heightened scrutiny of their TCO. FreeStor users recently surveyed by IDC reported an average savings of $243,000 in annual storage costs, representing a total return on investment of 448%, and a payback period of 5.5 months. For hyper-scale environments, it's hard to imagine a more efficient way to control costs.