Nexenta is a software-only, software-defined storage (SDS) solution that enables users to create storage systems with the x86-based hardware of their choice and disk or flash-based arrays, with support recently added for all-flash arrays. Based on the ZFS file system, NexentaStor runs on VMware, Windows, Red Hat Enterprise Linux (RHEL) or Citrix, in a scale-up architecture supporting the NFS, CIFS, iSCSI and Fibre Channel protocols.
Software-defined, not software-based
During our meeting at VMWorld 2014, Nexenta made the distinction that most other products calling themselves “software-defined” are really “software-based”. According to the company, true software-defined storage solutions don’t include hardware but they must actually store data. In their view, SDS should change the business model of how storage is sold, not just enhance or optimize an existing storage infrastructure. In fact, they claim that most software-based storage systems really only benefit the storage vendor, not the user, and gave the following example.
Storage vendors have historically designed their systems with proprietary hardware and software, and charged relatively high margins for their technology. Lately, these vendors have incorporated SDS principles into their designs to create software-based storage, allowing them to replace their proprietary hardware with the same x86-based commodity systems that the industry has been migrating to.
Who benefits from commodity hardware?
This has dramatically lowered their costs of manufacturing, but these savings haven’t made it to the user, at least not in a significant way. These vendors are now using commodity hardware but are still making the kinds of margins they used to on proprietary hardware. Software-defined storage, like Nexenta, never had any ties to proprietary hardware or its margins, allowing users to capture the savings from a software only, software-defined storage solution.
Earlier this month, the company released NexentaEdge, a scale-out object storage solution that runs on Ubuntu and RHEL and supports iSCSI, Cinder, Swift and S3 APIs. The system provides global, in-line de-duplication and object-level replication, with file services and erasure coding on the roadmap. Object storage gives users a platform that’s ideal for the large, unstructured data sets that are stressing traditional storage infrastructures in many industry verticals like big data, cloud providers, rich media and even enterprises running their own private clouds.
NexentaConnect for VSAN is a product that integrates with VMware, providing NFS and SMB file services that VMware users can manage through vCenter. It also provides security and domain integration, plus data reduction via in-line compression and deduplication, caching and health monitoring.
Users can actually deploy the freeware “Community” edition of NexentaStor for up to 18TB of capacity or buy the enterprise license for larger deployments. Nexenta software is available through VARs and integrators and through Dell, which packages NexentaStor with server and storage hardware in a single SKU. SuperMicro is on the roadmap with a similar package and Nexenta provides reference architectures with HP and other hardware suppliers.
At Storage Switzerland, we have maintained a relatively loose definition of SDS instead of getting into an argument of what “real” SDS is. That said, if one takes a totally purist stance, solutions like Nexenta would be considered “true” SDS. This doesn’t mean that the ways other vendors have rolled it out, including what the company calls “software-based storage”, don’t have value. As is always the case, there are pros and cons to both.
Nexenta’s contention that storage vendors are misusing the term “software-defined storage” may seem like an issue of semantics, but their claim that some of these vendors are also overcharging for their products is one that users might want to look into. Part of Nexenta’s value is that it gives companies maximum hardware flexibility and provides the potential for compelling cost savings, based on the economics of commodity hardware.