Qumulo Debuts QaaS, a File Lake on the Azure Cloud
Organizations that want to store large amounts of file data in the cloud and access them via NFS and SMB protocol may be interested in QaaS, a hosted version of its data lake for files–a file lake–running in the Microsoft Azure cloud.
Qumulo’s distributed file system has been adopted by companies that need access to large amounts of data stored in files. That includes rendering films in media and entertainment, drug discovery in pharmaceuticals, and e-discovery in the legal profession.
Many of these companies want to get out of the business of owning the infrastructure that these large Qumulo clusters run on, says Ben Gitenstein, the company’s vice president of product management.
“They’re looking to get rid of their data centers,” Gitenstein says. “They want out of their data centers and they’re looking for large scale file storage in Azure so that they can move all their applications up.”
With QaaS—which stands for Qumulo on Azure as a Service—customers don’t have to worry at all about the underlying infrastructure. That’s the big attraction.
“On-prem, they think a ton about nodes and disk and what not because they have to,” he says. “But in QaaS in Azure, one of the central ideas of the service is to make this simpler and make you not to have to worry about nodes, for example. We want to take care of all of that for you.”
While customers have a range of data lake options available to them on Azure, QaaS is the first and only cloud offering petabyte-scale access to file-based data, Gitenstein says. Microsoft offers a file system with Azure Files and Azure NetApp Files, but it’s not something that is going to scale past 100 TB, Gitenstein says.
“Our customers need to be able to put billions of files and petabytes of data into one namespace,” he says. “Nobody accesses all of that at the same time. But they need to be able to access multi-terabyte or hundred terabyte-size working sets and do tens of gigabytes per second of throughout against it at any given time.”
These workloads start to blend into HPC territory. For example, autonomous car companies want to be able to access large amounts of image data to train their deep learning systems. Drug companies also want to search for patterns buried in genomic data emanating from DNA sequencers from Illumina, for example.
With QaaS, customers won’t get access to all the knobs and switches they would have at their disposal if they were hosting a Qumulo cluster themselves. That limits their ability to dial in the performance of their file lake. But Qumulo is focused on delivering a high-performance file service with QaaS, even if their customers aren’t exposed to all the underlying complexity. In fact, it’s designed specifically to shield them from it.
“In our world, they’re looking for tens of GB per second of performance or less and they’re looking for hundreds of thousands of IOPS, sometimes millions,” Gitenstein says. “We can serve that really easily and then we have lots of knobs on our side that we can play with to increase or decrease the amount of performance available.”
If the performance of the cloud is not up to par, then Qumulo can add more Azure VMs to the customers’ cluster, at their request. There are general performance SLAs and capacity SLAs, which are the two metrics that the QaaS offering is based on. It does not yet expose the performance SLA in dollar terms, but that may come in the future.
“We really try to get customers to not think much about nodes,” Gitenstein says. “That’s an on-prem kind of notion. So instead, we’re trying to get them to think of just the amount of capacity they need and the amount of performance they need.”
QaaS runs atop the Azure BLOB service, which is the underlying data store. Qumulo customers can define what Azure region they want their QaaS installation running in, in order to get the maximum performance out of their applications, which will typically be running in the same lake. The offering does not support multi-cloud or hybrid-cloud setups, but it does provide tools for moving and sharing data among different clusters.
There is substantial pent up demand to for customers to migrate their on-prem Qumulo customers to the Azure cloud, Gitenstein says. “I’ve had customer who have been beating me up for years saying, when can I have Qumulo on Azure?” he says. “[The customers] have deep partnerships with Azure. They’ve signed big agreements.”
As part of Qumulo’s partnership with Microsoft, customers can now spend their Azure financial commitments on QaaS. “That turns out to be really valuable,” Gitenstein says.
Pricing for QaaS is on par with Azure Files and NetApp Files, Gitenstein says. The service is only available in the US at this point, with additional Azure regions expected in the future. At some point, the company may also choose to offer similar services on other public clouds. For more info, see the company’s website at www.qumulo.com.