Most data centers give up on their storage consolidation projects. The storage mainframe approach described in blog 2 is too costly and too restrictive, and the traditional software-defined storage approach described in blog 3 doesn’t support broad enough use cases. As a result, most organizations are facing a storage sprawl epidemic. The typical data center has six or more storage systems from three or four different storage vendors.
Storage sprawl does allow the organization to deliver workload specific storage systems but the cost to purchase, manage and protect these different systems threatens to consume almost the entire IT budget. Instead of pursuing a workload-specific strategy IT should look at areas of consolidation that worked elsewhere in the data center and see if they can find a storage strategy that mimics those techniques.
Two examples of a successful storage consolidation strategy are server virtualization and software-defined networking. Both techniques create a common software layer via a hypervisor or a network operating system, yet still, enable the organization to select the hardware most appropriate for the task at hand. Although the hardware is different, the software layer enables IT to manage the server or network infrastructure from a single interface.
As discussed in blog 3, some software-defined storage (SDS) does enable the organization to manage different hardware, but in most cases, the software does not support all the storage use cases. Many storage software solutions only support a subset of a use case, like flash-only primary storage. A storage version of a hypervisor or network operating system needs to support block (iSCSI and Fibre), File (SMB and NFS), object storage and can instantiate itself in the cloud.
Covering all these protocol and location use cases enables the organization to be standardized on a single storage software solution while still having hardware flexibility. The software though needs to do more than provide hardware flexibility. Supporting different hardware is just the first step. The storage software solution also needs to take full advantage of the storage hardware system’s capabilities. Full exploitation of the storage hardware is not just a problem for software-defined storage it is also a challenge for turnkey storage system manufacturers.
The source of the problem is the efficiency of the storage software. The CPUs that power the modern-day storage system are now multi-core. Developers need to write their software in a way that they can fully multi-thread its execution so that the software uses multiple cores effectively.
Today’s NVMe SSD media offers very high performance and extremely low latency. These drives expose inefficient software. One has to look at the raw performance of an SSD, multiply that by the number of drives in a typical storage system and then see what the vendor is reporting for their performance. In most cases, it is less than 10% of the potential raw performance.
StorONE addresses all these challenges and delivers a storage software solution with the capability of genuinely consolidating storage. First, the software supports a wide variety of protocols, both on-premises and in the cloud. Second, it not only supports a wide variety of hardware it delivers much closer to the raw performance potential (as much as 90%) of those systems. StorONE delivers a double ROI. It uses commodity hardware like SDS solutions, but it also needs less of that hardware since it can deliver close to the raw performance of the system.
Finally, it is available in several business models, software only, turnkey system or cloud-like subscription, so the customer can purchase in the way most appropriate for their budget.