Datrium Announces DVX 3.0 with Split Provisioning

August 15, 2017 Off By David
Grazed from Datrium

Datrium announced Datrium DVX 3.0 with Split Provisioning, a fundamentally new and different method of scaling converged infrastructure for private clouds.  DVX with Split Provisioning is the industry’s first server-powered converged infrastructure system to fully separate scaling of host storage speed and persistent capacity to simply and incrementally match resources to evolving tenant requirements. The new software provides scaling from 1-128 Compute Nodes, or up to 200 gigabytes per second (GB/s) IO bandwidth, and from 1-10 Data Nodes, or up to 1.7 petabytes (PB) of capacity, in a single composable pool. Datrium will demonstrate its DVX platform with Split Provisioning at the VMworld conference in Las Vegas, Aug. 27-31, booth #618.

Unlike Open Converged Infrastructure, Hyperconverged infrastructure (HCI) approaches maintain persistent data on every server and are limited to 2 or 3 simultaneous server failures before users experience a data outage. As a result, HCI users tend toward smaller clusters, across which servers and workloads should be homogeneous. According to Gartner, "in today’s world, most hyperconverged deployments are in the range of 8 to 16 nodes.", so rackscale deployments need multiple clusters, each of which needs configuration and administration.  Gartner has concluded that "…many mainstream enterprise IT directors do not yet trust SDS in HCIS solutions to deliver multipetabyte capacity at scale for Tier 1 mixed workloads that require low latency.".

 

10X Greater Consolidation

With Split Provisioning, Datrium DVX scales compute with both primary and secondary storage resources in a radically new way. Compute Nodes run all workloads in local commodity flash for high performance and low latency. Each Compute Node writes persistent data to network-attached, capacity-optimized Data Nodes, which are similar to purpose-built backup appliances optimized for cost efficiency and sequential IO. Because Compute Nodes are stateless, any number of them can go down for service without affecting data availability.

Until now, Datrium DVX supported a single Data Node.  With DVX 3.0 and Split Provisioning, a single Datrium DVX can pool multiple Data Nodes, each of which adds capacity and write bandwidth. While the minimum DVX configuration is one compute node and one data node, Split Provisioning allows scalability up to 128 compute nodes and 10 data nodes in a single system.  Adding a Data Node to a pool is a one-command operation, after which the system automatically and non-disruptively rebalances capacity. System-wide performance and capacity increases up to:

  • 200 GB/s read bandwidth and 18 million input/output operations per second (IOPS), an order of magnitude higher than market-leading all flash arrays, ideal for read intensive applications like Microsoft SQL Server or virtual desktop infrastructure (VDI) boot storms;
  • 8 GB/s write throughput, ideal for write-intensive applications such as ingest operations for big data analytics or data warehousing;
  • 1.7 PB of effective capacity.

"At Parametric, we provide engineered portfolio solutions to institutional investors and private clients built on quantitative, rules-based analysis. Our largest private cloud environment has grown 200-300 percent over the last year, and have received requests for 80 to 100 terabytes for a single project," said Ben Garforth, Parametric Portfolio Systems Engineering Manager. "With Datrium and Split Provisioning, I can now handle our massive data growth cost effectively and non-disruptively within a single resource pool. And better still, I don’t need to manage a thing-I check in on the Datrium solution once a month and that’s about it."

"Datrium’s premise has always been that their "open convergence" method delivers better overall scalability, along with independent scalability of performance and capacity, compared to hyperconverged architectures – in essence, private clouds with virtually limitless performance and scale with the ease of public clouds" said Arun Taneja, Founder and Consulting Analyst of Taneja Group. "Split Provisioning delivers further proof point for their claim. With the pace of innovation coming from Datrium, I can’t wait to see what comes next."

Converging Compute with Primary and Secondary Storage

Datrium DVX converges compute with both primary and secondary storage, and Split Provisioning takes this to an entirely new scale. Compute Nodes write data to low cost, disk-based Data Nodes with always-on global deduplication, inline compression, erasure coding and Blanket Encryption. Data Cloud, Datrium’s built-in cloud data management software, can now manage up to 1.2 million snapshots per DVX, replicable to other DVX’s or to Amazon Web Services.

For additional data protection, DVX 3.0 now offers application-consistent snapshots for Microsoft Windows-based applications such as MS SQL Server. Datrium Volume Shadow Service Provider leverages instantaneous and scalable DVX Snapshots, eliminating VM stuns and related sluggish application performance for large data sets.

"As specialty appliances decline, the future of private clouds will look like DVX," said Brian Biles, Datrium CEO and Co-founder. "Split Provisioning enables an order of magnitude more application bandwidth than most all-flash arrays, much simpler rackscale consolidation than HCI, and better secondary storage than either.  This is tomorrow calling."