We talked with Exablox a month or so ago at Storage Field Day 10 (SFD10) and they discussed some of their unique storage solution and new software functionality. If you’re not familiar with Exablox they sell a OneBlox appliance with drive slots, but no data drives.
The OneBlox appliance provides a Linux based, scale-out, distributed object storage software with a file system in front of it. They support SMB and NFS access protocols and have inline deduplication, data compression and continuous snapshot capabilities. You supply the (SATA or SAS) drives, a bring your own drive (BYOD) storage offering.
Their OneSystem management solution is available on a subscription basis, which usually runs in the cloud as a web accessed service offering used to monitor and manage your Exablox cluster(s). However, for those customers that want it, OneSystem is also available as a Docker Container, where you can run it on any Docker compatible system.
Last year they introduced their 4312 OneBlox which has 10GbE links and slots for 12 disk drives. The 4312 supports any mix of capacity or performance disk drives (8TB max., 1TB min.)) and a minimum cluster is 1-Oneblox with 3 disk drives. Their max cluster (at the moment) is 7 nodes.
Immutable store with dedupe, compression and data protection
The object store backend provides an immutable data space which they can use to provide an almost continuous snapshots. Dedupe, compression and (continuous) snapshot can be turned on or off on a Share or Mount point basis. They stated that dedupe is a variable length offering, which provides great data reduction ratios for backup data.
We didn’t discuss this but I assume they use 3-way disk replication to protect against drive failures. In a minimum cluster, this insures that data is replicated across all 3 drives and with a 3 node cluster, they would replicate data across all the nodes to protect against node failure.
Recent enhancements have been made to their cross cluster data mirroring or storage replication. It used to be one directional only and only 1 to 1 (asynch) replication on a share/mount point basis. But with the latest release of OneBlox software they now support 1 to many, Many to 1 and Many to Many and they also now support bi-directional cross cluster (asynch) replication. Realize that data being replicated is always deduped/compressed objects (if enabled) and it’s on a per share/mount point basis, so there’s a lot of flexibility in how replication is configured and how much storage would be required at each site.
Economics of Exablox storage
Pricing for the 4312 OneBlox is $12K and current retail price for 8TB disk drives is something on the order of $440/each, so for less than $20K you can have 32TB raw and over 100TB with dedupe and compression turned on. In a 7 node/fully populated cluster we’re talking PB of deduped/compressed storage. Also the 4312 supports mixed capacity disk drives.
The $/GB cost of disk capacity is going down all the time, sometimes because new higher capacity drives are available but just do to volume production or competition. In any case, you buy OneBlox appliances today and then populate it with today’s disk drives, but tomorrow when you need more capacity, you go out and buy cheaper ($/GB) capacity and just plug that in. Riding the disk drive cost curve had been the realm of storage vendors and hyper-scale web properties in the past, but with Exablox, nowadays anyone can do this.
Exablox’s unique, bring your own disk model is sort of a cross between software defined storage, where you supply the hardware and the vendor supplies the storage software vs. appliance storage where the vendor supplies both the hardware and software.
It might be easier for a customer to adopt Exablox’s low-cost storage solution over a full software defined approach because they don’t have to supply server hardware and virtualization software. The customer gets to take advantage of a major portion of the cost savings (from disk capacity), while the storage vendor eliminates much of the cost and burden of hardware & OS qualification/validation required to support a full software defined storage approach.
This all assumes it’s much easier to validate and qualify every vendor’s current disk drives than it is to qualify every vendor’s virtualization OS, server and disk drive offerings. It seems to me there are a lot less disk drive vendors than server vendors and when you multiply this by virtualization providers OS levels it get’s much worse
CEO Doug Brockett, mentioned that they have over 300 customers after only 1.5-2 years in the market and have $52M in VC funding for the team. They mentioned that ~50% of current customers use Exablox for backup storage and the remaining use it for primary storage of unstructured data. Some use it for both.
With support for mixed capacity drives it seems like an easy storage solution for SMB customers. But with the new DR replication capabilities they seem to be going up market, more into the enterprise space with more sophisticated business continuance requirements…
Photo Credit(s): from their presentation, video online at Techfieldday.com