LeftHand refreshes iSCSI SAN software

Beth Pariseau

LeftHand Networks Inc. announced version 7.0 of its SANiQ iSCSI SAN software, which according to the company contains dozens of updates, including performance boosts, new data replication and failover options, and more efficient thin provisioning. The most popular updates with LeftHand users are the new features around geographically distributed clusters and replication for disaster recovery. In particular, the ability to have LeftHand's clustered storage nodes within the same logical system but operating on separate Ethernet subnets has users interested. "That's a big issue for us," said Bob Lamb, general manager of hosting services for Strategic Business Solutions Inc., a managed service provider and a LeftHand customer since February. Lamb said that being able to isolate domains belonging to different clients on shared storage will allow his company to pack in more clients on the same box without worrying about data security. More on iSCSI SAN Temp agency moves hundreds of terabytes to iSCSI SANIntransa launches 10 GigE IP SANColorado State picks EqualLogic over EMC for ExchangeISCSI brings VMware to a new audience Another user, Jimmy Reid, director of technical operations at the University of Maryland, said the new feature will help his organization plan out its disaster recovery scheme more easily. Another new feature that will play into that, Reid said, is the Multisite High Availability and Disaster Recovery (HA/DR) Solution Pack, an additional license that will allow four-way replication between geographically separate systems (previously the limit was two-way replication). "We can bring up four locations in different physical locations, manage them all centrally and get four-way replication without having to run fiber between the data centers." Other features highlighted by users revolve around more efficient use of capacity, including what LeftHand is calling "Thin Provisioning 2.0" and the ability to share unused direct attached storage (DAS) space on VMware Inc. ESX servers. Previously, LeftHand's thin provisioning worked by "stairstepping" additional capacity for the thin-provisioned volume when new data reached its previous threshold, growing the volume in preset chunks. This new version shows new allocation in smaller intervals closer to the actual allocation, resulting in a more efficient use of space, the company claims. The new software also adds more automation to provisioning, including automated detection of space and thresholds, rather than relying on asking the administrator to set levels at startup, automated alerting for capacity management and the recommendation of provisioning options based on LeftHand's historical experiences with other customers when alerts are issued. "We're constantly shifting volumes around and adjusting thresholds" when it comes to provisioning as it is now, Lamb said. "This will take several hours per month in management time off our plate." The ability to fold unused DAS on physical hosts running VMware, according to Reid, "is going to be huge for us. There's almost 2 terabytes (TB) that we can immediately start using that's sitting empty in our environment." More updates -- 'Self-healing,' performance, Windows Solution Pack In addition to the updates flagged by users, LeftHand is publicly announcing about a dozen more features. "Proactive Self-Healing" introduces a new data integrity check/disk scrubbing feature that reads every block of data on each of the LeftHand iSCSI SAN nodes once a month in the background. The goal is to identify bad read errors before they happen. If an error or bad sector is identified on a particular drive, LeftHand will restore it from a mirrored copy of the data on one of the other nodes in the system. If an error happens between checks, this same rebuild mechanism can be started manually to fix the error. Also in the background, the new software will perform a variation on its standard defragmentation algorithm to optimize volumes within the system, the better to free up more empty sequential pages for better overall performance. The company is claiming an increase in sequential throughput for both read and write operations by using larger stripes in its network RAID system, as well as performance improvements for snapshots, especially during the deletion and cleanup process. LeftHand said this is thanks to a new metadata layout within the software. A new software wizard, contained in a virtual machine residing on the LeftHand storage area network (SAN) and replicated with the rest of the data, guides administrators through the failover/failback process when recovering from snapshot backups at a secondary site. In addition to the Multisite HA/DR Solution Pack, LeftHand is coming out with a Windows Solution Pack that also requires a separate license and includes plug-ins for Virtual Disk Service (VDS), Volume Shadow Copy Service(VSS) and support for Windows Multipath I/O (MPIO). The company said it's still finalizing pricing on the HA/DR pack, but the Windows pack will cost $500 per server. Other updates are focused on simplifying management of the system. A global configuration feature allows users to create settings for users and groups once, and then all the nodes in the system will inherit the settings, including newly added nodes later on. Snapshots can now be recovered through the self-healing system even if there's a double error between overlapping snapshots. RAID 6 is now supported on Hewlett-Packard Co.'s (HP) DL320 server hardware (one of the packaging options with LeftHand's meet-in-the-channel sales program). The software now contains awareness of the hot-swap process after a disk failure, support for multiple iSCSI targets and sessions per volume, a "pause" and "resume" button for snapshot schedules and a more detailed reporting interface. Users and analysts want better application integration According to Tony Asaro, an analyst with the Enterprise Strategy Group (ESG), one area where LeftHand lags behind competition from legacy and emerging competitors alike is in its integration with popular applications, from VMware to Oracle and Exchange. "I'd like to see them integrate the way NetApp has with SnapManager, for example," he said. He added that LeftHand could also use its virtualized management layer to pack in more features down the road, including inline data deduplication for primary storage. The University of Maryland's Reid said application integration is also on his wish list for the product. "I'd like to see a common management interface for my LeftHand SAN and VMware."