Software Defined Storage (SDS) was supposed to displace legacy hardware storage solutions that locked customers into a particular vendor. Years after its introduction SDS continues to struggle to gain critical mass within data centers. Today most data centers, especially at the enterprise level, use solutions from vendors that provide a turnkey software and hardware package, even though most of these vendors claim to be “software vendors”. The problem is that most SDS solutions require IT to make too many compromises.
For “software to eat hardware” from a storage perspective, SDS needs to fulfill its promises of hardware flexibility, the abstraction of software advances from hardware advances, increased storage efficiency thanks to the pooling of storage resources, and ease migration of data between hardware platforms. SDS needs to deliver on these promises while at the same time not impacting performance. Instead, SDS should improve performance.
DataCore Software Delivering on the SDS Promise
DataCore Software has been in business since 1998. During its 20 years, it has remained focused on delivering the promise of a software-defined storage infrastructure that abstracts the software purchase decision from the hardware purchase decision. Additionally, DataCore doesn’t stop at just abstracting the software from the hardware since it also adds features to its solution, SANsymphony, to take advantage of its unique position in the storage infrastructure.
Most SDS solutions, for the price of hardware independence, come with a performance tax. However, DataCore improves performance. It does this in two ways. First, its Adaptive Parallel I/O feature brings real multi-threaded processing to the storage infrastructure enabling all CPU cores to work on I/O processing in parallel. Most systems process I/O in sequential order, i.e., serially, which forces applications to wait on I/O. Second, the SANsymphony software also uses RAM caching to further accelerate I/O. DataCore claims that customers see as much as a 5X increase in storage performance.
Providing Hardware Flexibility
DataCore’s solution manages virtually any storage hardware in almost any configuration. The storage systems can be SAN-based or directly attached to the systems running the DataCore software. The software can run on dedicated servers providing the storage controller function, or it can run in a hyperconverged configuration with the software running within the hypervisor cluster.
Abstracting the Storage Purchase Decision
As newer, more powerful servers and faster storage media come to market, the DataCore customer moves their software investment to these new platforms, instantly taking advantage of new technology without having to repurchase software or change operating procedures. Most vendor-provided turnkey systems force the organization rebuy the software they already have each time they upgrade hardware.
Pooling Storage Resources
DataCore takes advantage of its place within the storage infrastructure by pooling storage hardware resources. IT now manages different systems from different vendors, even white box systems, from a single interface. The single software controlling the various hardware resources means, for example, that taking a snapshot on vendor A’s hardware is done the same way as on vendor B’s hardware.
One of the aspects of DataCore that makes it interesting is that the software doesn’t merge the storage systems. They are still separate systems. But, IT manages them from a single interface with standard commands and software features. Consequently, IT can buy hardware for a specific purpose without increasing storage administration costs.
To manage data movement between these storage systems, DataCore automatically tiers data between these different hardware platforms at a block level. Auto-tiering enables the organization to optimize storage resources, lowering the amount of flash it needs to invest in because it can automatically tier to less expensive hard drive-based systems. It can even tier to cloud storage if the customer so chooses. The net impact is a dramatic reduction in hardware storage investment costs.
The combination of automated data tiering and the abstraction of the storage software purchase also give DataCore an impressive migration capability. When IT implements new hardware into the storage infrastructure, the customer can immediately move specific datasets to it or enable auto-tiering to move data to it based on policy. If the organization purchases a high-performance flash system, IT can set it up as the primary tier for active IO. Auto-tiering migrates data to it based on the activity profile of the data, i.e., on how active the data is. Alternatively, if IT implements a high capacity storage system, another policy can automatically move less active data sets to it.
As a technology, SDS seems appealing. The problem is it has not executed well in the data center which explains the high percentage of enterprises that continue to buy vendor-specific turnkey solutions. DataCore is one of just a few SDS vendors that actually fulfill all the promises of SDS and even expand its potential.
DataCore exceeds the four essential elements of SDS. It provides hardware flexibility as well as configuration flexibility. It provides an abstraction of the software purchase from the hardware purchase, even making old hardware perform better. It increases storage efficiency by pooling storage resources and features lowering administrative costs. Finally, it eases data migration between storage hardware platforms.
Hardware lock-in makes even less sense today than it did twenty years ago when DataCore first came to market. New hardware advances like NVMe, Persistent Memory and increases in networking speeds, continue to become available. However, now those advances aren’t only available from a specific vendor like in the past. Instead, they are available from almost every hardware supplier at the same time. In this environment, portable, abstracted storage software makes more sense.