The company also added synchronous replication to RecoverPoint, 8 Gbps Fibre Channel (FC) and 10 Gigabit Ethernet (10 GbE) support in Clariion, and general availability of drive spin-down in Clariion arrays through EMC and partner Dell Inc.
VM-aware Navisphere improves storage management for virtual machines
To accommodate the proliferation of virtual machines, EMC refreshed the FLARE microcode for Clariion CX4 systems to allow the attachment of more initiators or endpoint connections. The CX4-960 – the largest Clariion array – can now support up to 4,096 initiators vs. 512 before this upgrade. The 10 GbE and 8 Gbps FC support gives the arrays more bandwidth as volumes of virtual machines grow.
Navisphere management console now allows customers to provision Clariion according to guest host rather than by physical server through an API integration with VMware vCenter Server. The vCenter integration gives customers a view of the back-end disks, RAID groups, storage groups, LUNs and volumes relating to a guest host and its storage objects.
Michael Passe, storage architect at Beth Israel Deaconess Medical Center in Boston, said he'll look into this feature, but he has some technical questions. "If it gave us some view into the guests or allowed provisioning to the guest rather than the ESX physical [host] we'd like that, [since it's] probably easier for troubleshooting I/O contention," he said. EMC said N_Port ID Virtualization (NPIV) isn't required at the host bus adapter (HBA) level to make VM-aware Navisphere work, but Passe said he wants to look under the hood to see exactly how that works before deciding to deploy the feature. "The interesting question is, 'How do you provision only to [one] guest if they share the same HBAs without NPIV?'" he asked.
An EMC spokesperson said a technical whitepaper is being drafted on how "users can mix VMs with NPIV enabled and those without [NPIV] on the same physical HBA or port" within Clariion, but that the information wasn't yet being made public.
VM-aware RecoverPoint and Replication Manager
The new integration of RecoverPoint and Replication Manager with VMware vCenter Server means customers can now see the replication status of each virtual machine in the RecoverPoint interface; in addition, Replication Manager supports image-based recovery of virtual machines and data. Chad Sakac, EMC's vice president who oversees the VMware technical alliance, said RecoverPoint's new synchronous replication support will maximize the recovery point objective (RPO) for customers who use RecoverPoint with VMware's Site Recovery Manager (SRM) for disaster recovery.
The combined updates to RecoverPoint and Replication Manager allow the restoration of a single virtual machine from either a file or block interface, Sakac said.
Spin-down now shipping
The drive spin-down for CX4 arrays that EMC announced last August is finally becoming generally available for EMC and Dell customers.
Spin-down can be set according to policy at either the array or RAID set level on SATA disks only, according to Brett Roscoe, Dell's senior product manager for enterprise storage. Once a RAID set gets an I/O request, it spins back up again. The time it takes to respond to an I/O request varies based on the number of spindles. EMC tested the largest size RAID group of 15 drives and found it responded in 54 seconds if the drives have been spun down.
"So far, EMC applications have been able to respond to that latency," Roscoe said. "We're working on best practices and support matrices to make sure people don't have issues with latency and I/O." Dell will also add storage consulting and assessment services for EMC products to go along with the ones it already offers for EqualLogic customers.
10 GbE, FCoE and shifting markets
Dell is adding 10 Gigabit Ethernet support in CX4 before its EqualLogic iSCSI product line supports what some in the industry believe is the future of data center networking. Roscoe said he expects EqualLogic to support 10 GbE before the end of the year. Although rival NetApp recently added native Fibre Channel over Ethernet (FCoE) support for its storage arrays, EMC executives say native FCoE support won't become available until next year. In the meantime, debate continues over whether users will choose 8 Gbps or 16 Gbps FC, FCoE, or 10 GbE iSCSI and NAS.
Mark Peters, an analyst at Milford, Mass.-based Enterprise Strategy Group, said Clariion and other recent midrange upgrades from vendors such as BlueArc Corp./Hitachi Data Systems, Isilon Systems Inc. and 3PAR Inc. continue to blur lines between midrange and enterprise product categories.
"There's a lot of similar language in these announcements about 'enterprise-class features at midrange cost,'" Peters said. "It used to be that if you bought a smaller array, you didn't get as many features or the same reliability. Now it seems that we're moving to a market where 'midrange' means scale rather than the set of features or reliability you get."