Microsoft is adding new features to an upcoming version of Windows Server that are aimed at helping customers provide enterprise data storage services with commodity hardware.
Members of Microsoft’s high availability and storage group—Claus Joergensen, a principal program manager, and Ned Pyle, a senior program manager—appeared on the company’s online video series, The Edge, to preview some of the storage technologies being baked into the next major release of the server operating system. The new features, said Joergensen, are part of Microsoft’s continued pursuit of a software-defined approach to storage for the Windows Server ecosystem.
Following up Microsoft’s work on Windows Server 2012, particularly the SMB3 protocol and Storage Spaces, the latter of “which allows you to take commodity storage hardware components and create storage pools and virtual disks,” the company is doubling down on software-defined storage, according to Joergensen.
“Now, in the next release, we are continuing those investments,” said Joergensen. He also teased “new, exciting hardware that we will be embracing.”
Joergensen described software-defined storage as “super important” for the software giant, allowing the company “to increasingly remove the need for specialty hardware.” For customers, it opens the door to enterprise-grade data storage capabilities and management at a lower price tag.
“So you can take standard, off-the-shelf server hardware and you can build [a] high-availability storage solution as well a high-availability compute solution without the need for traditional specialty hardware,” he continued.
The new Shared-Nothing Storage Spaces feature lowers cost by creating pools of capacity and failover clusters that leverage direct-attached storage, bypassing the need for storage area networks (SANs). One feature Joergensen said he was “super excited about” is automatic rebalancing.
Customers will be able to add new nodes and allow Windows Server to self-optimize the data placement if, for example, they need to upgrade from a four-node set-up to six nodes.
With the new software, “we can rebalance that data across those additional two nodes so that you can consume all of it and get the additional performance and capacity without necessarily needing to create a new virtual disk and deploy new VMs [virtual machines] to realize it,” Joergensen said.
Microsoft is also seeking to minimize the potential for data loss with its new synchronous replication solution, Storage Replica, a step up from Hyper-V Replica, the company’s current asynchronous data protection scheme. With Hyper-V Replica, data can fall through the cracks if a mishap takes place between preset data mirroring intervals.
Storage Replica “is designed to fill in that gap that we’ve had in our storage stack the entire time,” said Pyle. “You can run everything Windows if you wanted,” he said of Hyper-V Replica, but customers were still required “to buy somebody else’s product if you wanted to do synchronous replication, or what we call, block replication.” The new solution is “purely agnostic to the hardware, agnostic to the storage, agnostic to the file system” and provides around-the-clock protection, he said.