What’s new in FLARE 32?

We recently installed a new VNX5700 at the end of July and EMC was on-site to install it the day after the general release of FLARE 32. We had planned our pool configuration around this release, so the timing couldn’t have been more perfect. After running the new release for about a month now it’s proven to be rock solid, and to date no critical security patches have been released that we needed to apply.

The most notable new feature for us was the addition of mixed RAID types within the same pool.  We can finally use RAID6 for the large NL-SAS drives in the pool and not have to make the entire pool RAID6.  There also are several new performance enhancements that should make a big impact, including load balancing within a tier and pool rebalancing.

Below is an overview of the new features in the initial release (05.32.000.5.006).

· VNX Block to Unified Services Plug-n-play: This feature allows customers to perform Block to Unified upgrades.

· Support for In-Family upgrades: This feature allows for the In-family Data In Place (DIP) Conversion kits that are to be available for the VNX OE for File 7.1 and VNX OE for Block 05.32 release. In-family DIP conversions across the VNX family will re-use the installed SAS DAEs with associated drives and SLICs

· Windows 2008 R2: Branch Cache Support: This feature provides support for the branch cache feature in Windows 2008 R2. This feature allows a Windows client to cache a file locally in one of the branch office servers and then use the cached copy whenever the same file is being requested. The VNX for File CIFS Server operates as the Central Office Content Server in support of the Branch Cache feature.

· VAAI for NFS: Snaps of snaps: Supports snap of snap of VMDK files on a NFS file systems to at least 1 level of depth (source to snap to snap). Though the functionality will initially only be available through the VAAI for NFS interface, it may in future be exposed through the VSI plug-in as well. This feature allows VMware View 5.0 to use VAAI for NFS on a VNX system. This feature also requires that file systems be enabled during creation for “VMware VAAI nested clone support”, and the installation of the EMCNasPlugin-1-0.10.zip on the ESX servers.”

· Load balance within a tier: This feature allows for redistribution of slices across all drives in a given tier of a pool to improve performance. This also includes proactive load balancing, such that slices are relocated from one RAID group in a tier to another, based on activity level, to balance the load within the tier. These relocations will occur within the user-defined tiering relocation window.

· Improve block compression performance: This feature provides for improved block compression performance. This includes increasing speed of compression operations, decreasing storage system impact, and decreasing host response time impact.

· Deeper File Compression Algorithm: This feature provides an option to utilize deeper compression algorithm for greater capacity savings, when using file-level compression. This option can be leveraged by 3rd party application servers such as the FileMover Appliance Server, based on data types (i.e. per metadata definitions) that are best suited for the deeper compression algorithm).

· Rebalance when adding drives to Pools: This feature provides for the redistribution of slices across all drives in a newly expanded pool to improve performance.

· Conversion of DLU to TLU and back, when enabling Block Compression: This feature provides an internal mechanism (not user-invoked), when enabling Compression on a thick pool LUN, that would result in an in-place conversion from Thick (“direct”) pool LUNs to Thin Pool LUNs, rather than performing a migration. Additionally, for LUNs that were originally Thick and then converted to Thin, it provides an internal mechanism, upon disabling compression, to convert the LUNs back to Thick, without requiring a user-invoked migration.

· Mixed RAID Types in Pools: This feature allows a user to define RAID types per storage tier in pool, rather than requiring a single RAID type across all drives in a single pool.

· Improved TLU performance, no worse than 115% of a FLU: This feature provides improved TLU performance. This includes decreasing host response time impact and potentially decreasing storage system impact.

· Distinguished Compression Capacity Savings: This feature provides a display of the capacity savings due to compression. This display will inform the user of the savings due to compression, distinct from the savings due to thin provisioning. The benefit for the user is that he can determine the incremental benefit of using compression. This is necessary because there is currently a performance impact when using Compression, so users need to be able to make a cost/benefit analysis.

· Additional Tiering Policies: This feature provides additional tiering options in pools: namely, “Start High, then Auto-tier”. When the user selects this policy for a given LUN, the initial allocation is on highest tier, and subsequent tiering is based on activity.

· Additional RAID Options in Pools: This feature provides 2 additional RAID options in pools, for better efficiency: 8+1 for RAID 5, and 14+2 for RAID 6. These options will be available for new pools. These options will be available in both the GUI and the CLI.

· E-Trace Enhancements: Top files per fs and other stats: This feature allows the customer to identify the top files in a file system or quota tree. The files can be identified by pathnames instead of ids.

· Support VNX Snapshots in Unisphere Quality of Service Manager: This feature provides Unisphere Quality of Service Manager (UQM) support for both the source LUN and the snapshot LUN introduced by the VNX Snapshots feature.

· Support new VNX Features in Unisphere Analyzer: This feature provides support for all new VNX features in Unisphere Analyzer, including but not limited to VNX Snapshots and 64-bit counters.

· Unified Network Services: This feature provides several enhancements that will improve user experience. The various enhancements delivered by UNS in VNX OE for File 7.1 and VNX OE for Block 05.32 release are as follows:

· Continuous Monitoring: This feature provides the ability to monitor the vital statistics of a system continuously (out-of-box) and then take appropriate action when specific conditions are detected. The user can specify multiple statistical counters to be monitored – the default counters that will be monitored are CPU utilization, memory utilization and NFS IO latency on all file systems. The conditions when an event would be raised can also be specified by the user in terms of a threshold value and time interval during which the threshold will need to be exceeded for each statistical counter being monitored. When an event is raised, the system can perform any number of actions – possible choices are log the event, start detailed correlated statistics collection for a specified time period, send email or send a SNMP trap.

· Unisphere customization for VNX: This feature provides the addition of custom references within Unisphere Software (VNX) via editable source files for product documentation and packaging, custom badges and nameplates.

· VNX Snapshots: This feature provides VNX Snapshots (a.k.a write-in-place pointer-based snapshots) that in their initial release will support Block LUNs only and require pool-based LUNs. VNX Snapshots will support File Systems in a later release. The LUNs referred to below are pool-based LUNs (thick LUNs and Thin LUNs.)

· NDMP V4 IPv6 extension: This feature provides support for the Symantec, NetApp, and EMC authored and approved NDMP v4 IPv6 extension in order to back up using NDMP 2-way and 3-way in IPv6 networked environments.

· NDMP Access Time: This feature provides the last access time (atime). In prior releases, this was not retained during an NDMPCopy and was thus set to the time of migration. So, after a migration, the customer lost the ability to archive “cold” or “inactive” data. This feature adds an optional NDMP variable (RETAIN_ATIME=y/n, the default being ‘n’) which if set includes the atime in the NDMP data stream so that it can be restored properly on the destination.

· SRDF Interoperability for Control Station: This feature provides SRDF (Symmetrix Remote Data Facility) with the ability to manage the failover process between local and remote VNX Gateways. VNX Gateway needs a way to give up control of the failover and failback to an external entity and to suppress the initiation of these processes from within the Gateway.

 

Advertisements

Leave a Reply