Scale Computing broadens VM DR with HyperCore update

The latest iteration of Scale's hyper-convergence enhances cloning with block-level replication of VMs for failover and failback across clustered nodes.

Hyper-converged vendor Scale Computing Tuesday upgraded its HyperCore storage operating software to support disaster recovery across clustered HC3 nodes.

Version 6 of HyperCore includes multisite replication of virtual machines (VMs) and a browser-based management interface for tagging and filtering VMs in groups.

The new release supports failover and failback of VMs between local storage and remote backup on Scale's HC3 hyper-convergence platform. Automated local and manual snapshots are replicated remotely. HyperCore 6 tracks and ships only the delta blocks made from the most recent snap.

Replication can take place on either a per-VM basis or as VM groups tagged for specific databases or application workloads. Customers can clone the snapshot on the target cluster for manual failover. HyperCore 6 includes an HTML5-based graphical user interface to manage and monitor virtual storage consumption.

The HC3 platform combines compute, network, storage and virtualization in one box. Scale markets the 1U HC3 appliances in a minimum three-node cluster, aiming it primarily at small and medium-sized businesses. HyperCore software intelligence installs on commodity hardware running SAS and SATA hard drives, although Scale said hybrid nodes that include flash storage are on the product roadmap.

HyperCore 6 automates provisioning, managing KVM storage

HC3 systems use the open source Linux Kernel-based Virtual Machine (KVM) as the core hypervisor. HyperCore automated management software is a bare-metal hypervisor written around the KVM software kernel.

The HC3 includes LAN and backplane ports for networking. The HyperCore architecture relies on clustered components rather than a centralized management server and RAID to provide consensus on network outages and spin up backup storage accordingly. 

For HyperCore 6, the vendor reworked its Scale Computing Reliable Independent Block Engine (Scribe) to boost continuous VM-level protection. Scribe is a clustered block-storage software layer designed for direct consumption by KVM. It aggregates storage on block devices across HC3 nodes and presents it as a managed pool for load balancing and redundancy.

Scribe now takes VM snapshots and replicates block-level differentials to secondary storage, building on the cloning technology added in HyperCore 5 for instantly promoting writable snapshots to production storage.

Scalability expanded to thousands of VMs per node

Previous iterations of HyperCore supported about 50 VMs per node. Scale Computing claims HyperCore 6 can manage several thousand VMs per node.

Customers use point-and-click drop-down menus to create a VM, assign its IP address and initialize it on an HC3 cluster. HyperCore automatically loads software drivers and allocates CPU cores and memory to storage.

New VMs use Secure Socket Shell (SSH) protocols to create multiple tunnels to connect with other nodes in the HC3 cluster. The SSH tunnels transport block-level snapshot changes that are encrypted and compressed between primary storage and disaster recovery (DR) targets. The process works in reverse for failback, with delta blocks replicated from a DR site to a production data center.

Among hyper-converged vendors, only Scale Computing and Nimboxx are designed solely for KVM, although Nutanix supports KVM along with VMware and Microsoft Hyper-V. Open source software vendor Citrix moved into hyper-converged storage in January with the acquisition of scale-out storage software vendor Sanbolic.

Scale Computing CTO Jason Collier said the mobile GUI provides the foundation for subsequent HyperCore releases.

"Our core design tenets are simplicity, high availability and scalability, with simplicity above all," he said. "Breakdowns don't always occur nine to five, so we wanted to enable an IT admin to take corrective action from a smartphone at their kid's soccer game if necessary. Our goal is to make storage virtualization simpler for IT shops that have to be a jack-of-all-trades."

Collier said future releases could enable customers to use Scale nodes to create hybrid clouds for replicating VMs for archiving as well as recovery. Also on the drawing board is a software component for using Scale HC3 as an object data store in conjunction with other vendors' storage systems.

Next Steps

2015 poised to see expansion of hyper-converged market

Scale Computing receives Impact Award for SMB focus

Dig Deeper on Hyper-Converged Vendors and Products