LAS VEGAS — WekaIO, which emerged from stealth this summer with its high performance parallel scale-out file storage solution, has announced their WekaIO Version 3.1 release. The big news, just in time for AWS re:Invent here, is an enhanced snapshot functionality, which allows for an entire file system – plus data – to be backed up to Amazon S3.
WekaIO’s support for AWS itself is not new, as cloud storage is a fundamental part of their value proposition. Their solution provides scale out file storage for both public clouds and on-prem servers. It uses the compute layer for storage, aggregating all the local SSDs inside the servers into a single cluster, which is all natively managed as a single namespace inside the compute layer.
The company has, however, been fine-tuning its go-to-market strategy.
“We launched in July and have been very active since then, in terms of getting our messaging out,” said Barbara Murphy, WekaIO’s VP of Marketing. “We learned very early on that while people love the concept of software defined storage, the pragmatic reality is that they buy systems. As a result, we have been very active on the channel side, building up partnerships and routes to market.”
A key element in this is a new WekaIO partnership with HPE, announced two weeks ago.
“They will be reselling us,” Murphy said. “We are part of the HPE Complete program, but also engaged at a deeper level around the HPC group. HPE does not have a nice consumable way to address the high performance scale-out market. IBM Spectrum Scale is the leader in that space, but there hasn’t been another viable solution. Because we are a high-performance file system solution, we enable a much higher performance solution for HPE than they had. We are coselling together into that space, around machine leaning, and already have a joint design win in the rendering space in media and entertainment. HPE can also leverage other portfolio pieces, like their relationship with Scality, to have a joint solution with both a high-performance tier and a data lake, which is much broader than either Scality or us on our own.”
WekaIO is also collaborating with HPE and Mellanox, the interconnect vendor, to include native Infinband support.
“Infiniband is the interconnect of choice around HPC, which gives us a strong performance play in GPU intensive workloads,” Murphy said.
WekaIO’s Matrix 3.1 release and the deeper AWS integration is also intended to broaden WekaIO’s route to market by partnering with a dominant player.
“The AWS relationship gives us access to a global market, and makes our product consumable in an easy way, globally,” Murphy said.
The core enhancements in 3.1 are specific to Amazon S3. It supports cloud bursting to allow elastic growth in the cloud in response to peak workload periods, with a a Pause/Resume feature for S3. That feature lets you run on Amazon EC2 instances for high application performance, stop instances when peak demand fades, pay only for cost-effective Amazon S3 storage, then re-spin the Amazon EC2 instances when you choose.
“It will allow you to snapshot an entire file system to park it in S3 as low-cost storage until you need the file system again, and then you can rehydrate it at a later date,” Murphy said. “We can also allow you to snapshot to AWS from an on-prem location, giving much more elasticity to get compute resources on demand. We had snapshotting before, but it was snapshotting of files. This release lets you snapshot the entire file system – plus all of the data – to a remote location, which is a lot more powerful than just snapshotting the files themselves.”
Murphy referred to this capability as “a poor man’s disaster recovery strategy” – in the sense that it provides what is tantamount to a complete DR capability, without the need to purchase separate backup and DR tools.
“Remote backup to the cloud this way is also a simple way to check that backups work, and you can rehydrate in a very small cluster in the cloud,” she said.
Version 3.1 also has a couple of other advantages for HPC inside the AWS cloud.
“The only option for HPC until this has been Lustre [open source file system],” Murphy said. “The other nice thing is proactively working on a tight integration between GPU clusters, so it can provide a very high-performance tier into the GPU environment.
“We have also built a configurator tool to onboard, where you can specify what you want in terms of performance,” she added. “It comes up with the lowest cost solution for different iterations of pricing, with a single push of a button.
Murphy said that in the short time since 3.1 was announced, they have been getting a lot of direct inbound inquiries.
“We also announced a relationship with San Diego Supercomputing, an HPC center with advanced engineering, analytics and genomics capabilities,” she stated.
WekaIO will be demonstrating Matrix Version 3.1 at Booth #507 at AWS re:Invent 2017 in Las Vegas.