Happy day for EMC unified customers. A whole bunch of new integration, additional cost savings – all for existing and new EMC customers. Oh, and it’s all free :-)
Here’s the PR, but in usual fashion, I tend to like the nerd version.
Read on for more!
So – without further ado, what’s new and GA?
- Celerra F-RDE (file redundant data elimination) has been expanded and now reduces the amount of storage consumed by VMDKs when stored on NFS. In practice, this saves between 40-60% of capacity on top of thin provisioning.
- VM-level snapshot and clones which can hardware accelerate VM snapshot and clones when those are stored on NFS datastores. Of course, this is a reflection of the fact that the Celerra can do file-level snapshot and file clones (even across different filesystems!). This function is delivered via an extension of the vCenter GUI.
- Dramatically simplified provisioning in the NFS use case. You can easily create NFS datastores in a single screen – directly in the vCenter context. This configures the filesystem on the Celerra, and all the ESX hosts in the cluster. It automatically configures all the critical Celerra and ESX side parameters. You can easily also use it to reconfigure the datastore, as well as examine actual storage consumption.
- the Celerra SRM plugin for automated failback has been extended and now covers NFS (hey, sometimes the other guy is first, sometimes you’re first :-)
Here’s a demo, which shows it all in action.
Things to know:
- A ZIP file containing the GA release (1.0.7) of the plug-in and the release notes will be posted on Powerlink today.
- Find it here: Home > Support > Software Downloads and Licensing > Downloads A-B > Adapters for Third-Party Applications
- A Solution Guide has been posted on the Powerlink VMware Support Page.
- Find it here: Home > Support > Product and Diagnostic Tools > VMware Support > Celerra
- The plug-in requires DART 184.108.40.2061 or above. Refer to the release notes for more details.
- Internal EMC Users can obtain the DART 220.127.116.111 code on March 10th from this location:
Home > Support > Software Downloads and Licensing > Downloads C > Celerra Software
- Customers and Partners should follow standard procedures for upgrading to the new version of DART. Refer to the release notes for more details.
- Yes, I know the next question – does this work on the Celerra VSA? Not the currently available one (which is based on DART 18.104.22.168), but the Celerra VSA crew is committed to staying close to GA code rev, and the target release of the updated VSA (which continues to use less memory/CPU and run faster too!) will be out before the end of March. (UPDATE – commit made, commit delivered. download the 22.214.171.1241 version of the Celerra VSA here)
- For people who are familiar with NetApp – one bug to report that affects this and the NetApp RCU tool (including RCU 3.0) which extends the vCenter GUI in a similar way. The right click menu randomly disappears while operating the plug-in. The user has to restart the vSphere/VI client to get the menu back. It’s minor, and an annoyance. The good news (for both EMC and NetApp customers) in the development of the tool, we worked with VMware on a fix (that also happens to fix it with RCU), which is scheduled for the vCenter release train :-)
So – how does the data savings work?
- Celerra F-RDE does deduplication of file-level objects and compresses sub-file elements. Don’t think of the compression as a “zip” – rather it’s a real-time, very, very rapid compress/decompress.
- It is a post-process and background process – not an inline dedupe and compress.
- In the first version, it eliminated identical files, and then compressed files that were smaller than 200MB. This has been out for a while now, and used by boatloads of customers. Against generic NAS use cases, it’s awesome – adding between 30-50% capacity efficiency. The generic NAS use case is also one where capacity and performance are not linked (in other words there is not a relationship between the number of file objects, or the capacity required, and the number of CIFS OPs/Sec or NFS OPs/Sec). This did virtually nothing for the VDMK on NFS use case, as the big capacity consumers were large VMDK files.
- In the current version, the compression has been expanded to hit the VMDK use case. The way this works is that when a file is marked to be compressed (which can be done on a VM-level or file basis, or datastore or filesystem basis), writes are written to a small set-aside area – and then as the background process optimizes. Reads decompress the portion of the file is read. Again – I want to set expectations properly. This has shown 40%-60% additional space savings on top of a Windows Server 2008 VM, and has no performance impact during write operations, and about a 10% performance impact on read operations. Frankly – with that tradeoff, I think people will just leave this on.
While every approach has it’s advantages, it also has disadvantages. One advantage with this approach is we were able to do it with no impact to any other Celerra function, or limiting the filesystem in any way. It’s also not a function of the Celerra platform – so all, in each member of the family work the same.
With the Celerra and CLARiiON sharing more and more (lots in the iSCSI stack, and also the CBFS layer that handles thin provisioning and is also where this compression occurs) – you can see where this is going next :-)
EMC’s NAS+Block platforms are our fastest growing platform at EMC – enjoying it’s 11th quarter of consecutive double digit growth, and according to IDC, EMC now has north of 50% of the NAS market, more than double the nearest competitor according to their analysis. Today’s announcement fills some holes, extends some leads – in general a good thing for our customers!