EMC World 2010 : What’s New for the Celerra

Written By Joel Ramirez.

I attended the Celerra FAST Uncovered: Cloud Integration and Storage Efficiency session at EMC World this afternoon and, while the name is a bit of a misnomer for the session in two out of three ways, the veteran unified storage platform that is the Celerra is now coming full circle with storage efficiency.

First off, the Unisphere GUI is great.  It’s a long time coming and rest assured it can do everything that we’ve been forced to do separately with Celerra Manager and Navisphere Manager.  In fact, Recoverpoint management is built into it as well.  Now you can set quotas for a filesystem, expand a meta-LUN, and build a consistency group via one management interface.

The session really didn’t speak to the cloud integration piece at all.  Perhaps it’s the fact that fully automated storage tiering, or FAST, has such an integral place in the private cloud.  A fellow skeptic and I hammered away at the product engineer to test the relevance of the improved feature set of the Celerra.  The idea of FAST is to change the entire storage allocation paradigm.  FAST represents a pool of disk that can contain SATA, FC, and EFD.  It shuffles 1GB blocks of data among the different tiers based upon user policy and the Celerra’s continuous measurement of data access.  Hot spots are accommodated, for example, by moving the data to EFD’s.  The window of measurement is continually the last hour so the Celerra is using the latest data to determine where to position data among the different tiers of storage. 

The EMC Engineering team has an amazing tool to help scope a Celerra using FAST called Tier Advisor which will unfortunately only be available to EMC Internal.  This tool is being engineered to ingest .nar files, the performance logs of the Celerra, to gauge what percentage of disk is doing the bulk of the work.  It then allows the engineer or consultant to apply different scenarios to determine their impact and dependencies.  It gives information such as the change in power consumption, the number of disks and what types required to meet the change in scope, and it provides a graphical representation of the difference between the status quo and the hypothetical input. 

Finally, Flash Cache is a really exciting technology.  It allows you to expand the cache in a Celerra or Clariion with enterprise flash drives.  This is great news for companies who have to buy bigger Clariions that have higher cache capacities just because they need to feed their applications with bursty performance.  Instead they can throw a pair of EFD’s in a CX4-120 and have up to 200GB of cache!  By using solid state technology and simply adding more capacity to the fastest part of the array architecture, you can worry less, if at all, about spinning disk being the bottleneck.  From here on out, I’m all for CX4-120’s with a pair of EFD’s. 

Some other notes:

• Celerra Unisphere is a free upgrade if you are under software warranty (you need a CX4) and will manage older Clariion arrays, too.
• Compression is available in the Celerra as a pay-for feature, a good fit for less frequently accessed data on file systems (use with discretion as it it resource intensive)

• Virtual Provisioning will become free

• Auto-tiering, the meat of the FAST technology, is a pay-for feature
• Zero Space Reclaim is being introduced and will be a pay-for feature