Exablox Briefing Note
Despite the continual decline in the cost of flash, all-flash arrays remain out of reach for many medium sized data centers. They are left trying to band-aid performance problems with server side caching or hybrid arrays instead of fixing the performance problem once and for all with an all-flash array. If they make the jump to a small flash array, many believe they have to choose a solution that won’t scale as their business scales. Medium sized organizations have applications that can benefit from an all-flash array’s performance, but industry is lacking the solution they need.
The Requirements for a Tier 2 Data Center All-Flash Array
For mid-sized data centers to even consider an all-flash array it has to meet certain criteria. First, it has to be something affordable from day one. Most vendors require dozens of TBs of investment before the cost per GB of flash approaches the cost per GB of hard disk drives. The mid-tier data center does not need that kind of capacity, so the math never works out. They need a system that can start very small but scale to keep pace with organizational growth.
Second, the mid-tier all-flash array needs to provide the right level of performance. While these data centers have workloads that need more than hard disk drive performance, they don’t typically have a demand for millions of IOPS, or frankly, even hundreds of thousands of IOPS. They will, however, have multiple workloads that need 10k to 25k IOPS. So they need a system that can provide scalable performance over dozens of workloads.
Enter The Exablox Oneblox 5210 – All-Flash for the Mid-Tier Data Center
The Exablox 5210 is an all-flash array for the mid-tier data center. The system delivers about 50K IOPS per node at less than .50 cents per GB. It meets the first requirement of being affordable. The system can scale-out, up to seven nodes, but can start with as little as one. Also, the customer can supply its own SSDs. Each node supports 10 2.5″ drives. Using 4TB drives (8TB and 16TB support is coming) means each node can deliver up to 40TBs of raw capacity. Exablox includes deduplication and compression technology, which means each node should deliver an effective capacity of 200TB (assuming 5:1 data efficiency).
The scale-out architecture means scaling is available as the organization grows. The Exablox architecture allows for the addition of seven nodes the cluster. Storage automatically spans across pools, making the effective capacity 1.4 PBs, which should last the mid-tier data center perfectly.
Exablox has the feature sets you’d expect in primary storage including snapshots, replication and the aforementioned deduplication and compression.
5210 Use Cases
Exablox builds its systems on top of an object store, so scalability is in its DNA, along with the ability to manage unstructured data. The object store underpinning is carefully hidden from administration and the users. It looks and acts like an NFS or SMB file server. Access to storage is SMB and NFS, making the system excellent for virtualized environments and high performance file storage.
For larger enterprises, the 5210 could also be an ideal backup target. Many of these organizations are already investing in all-flash arrays and are investing in backup software that provides recovery in place capabilities. The problem is there is a disconnect, recovering on a slow disk backup appliance after the users have become accustomed with all-flash performance results in user dissatisfaction. The 5210 could solve that problem.
Mid-tier data centers need flash, and enterprises need a backup target that more closely matches their current flash investment. The problem is that the solutions targeting the mid-tier data center are lacking. Exablox’s 5210 hits all the mid-tier data center’s flash requirements and adds some capabilities they probably weren’t expecting. For enterprises the 5210 may be the ultimate flash backup target.