EMC World 2009: Day 2 recap. It's raining again...
So, what happened here? Well, there were two main topics that are both related to the new systems that was introduced a while back by EMC.
There was a session on the changes in the new Symmetrix Enginuity version 5874 for the new V-Max system. We were informed that among others the RAID-engine was redesigned. The new version allows for only 1 mirror per device instead of the previous 4 mirrors per device. They now allow all of the features on all of the RAID-levels, and they introduced the possibility of VLUN migration.
Also new was the support for large volumes. The limit increased from 16GB to 240GB. This should help a lot of people who were required to create large metasets before. Also, you can now create 512 hypers per drive.
You will find some changes in the management and provisioning section for this new version of Enginuity. Dynamic provisioning received some enhancements that reduce the number of steps required to present your storage to your host. You now create a storage group, a port group and a device group or initiator group. After that you are basically done. Create some templates once these prerequisites are created and you can present your storage with relative ease. People will also find that bin file changes are speeded up on this new release.
SRDF also received some upgraded numbers. The maximum number of SRDF groups is now at 256 groups per array, and 64 groups per FA-port. Still, you might want to reconsider splitting up all groups on 4 FA-ports.
They also introduced the new so called "DLDEV" or Diskless R2 Device, or in Symmetrix speech a so called R21 device that allows you to cascade to a 3rd site using SRDF/A. DLDEVs need to be on a V-Max though.
After an interruption I got in to a private V-Max session a bit later then I planned, but I still managed to get some new information.
Currently one of the less official recommendations that you will find is that if you have a large datacenter, you might consider distributing some engine groups in the various areas of your datacenter. Current distance limits you will find are at around 100 meters. Simple reason for this distance limitation is the fact that you are working with latencies. If you take the distance up to 100 meters, you will find that latency will increase with a factor 27. Upping the distance could create some unforeseen issues.
One of the questions was that with FAST we will see a tiering over the various storage media, but what will EMC do to tier my storage on the various storage classes. In a normal situation you want your mission critical data on high end arrays, and work your way down. One could imagine something similar to FAST on a different level that would allow you to better utilize the various availability levels of you storage hardware.
The engineers implied that they are working on this, but unfortunately would not go in to a great deal of detail.
One of the most heard questions so far is probably "When will FAST v2 be coming to the V-Max?". Current timeline indicated about 1 to 1.5 years, but no one was willing to talk about any of the technical details. This raises the question how far they actually are with the FAST technology, and that's a question that produces the same "no coment" remark and look. What they were able to tell though was that they are planning to introduce FAST on the Clariion series array, which will be a major money maker when they are able to deliver. Currently you will find around 300,000 CX's worldwide, and one can only imagine how many people might be interested in purchasing this technology, if and/or when it works. A first version of FAST v1 should be introduced at the end of the year according to Barry Burke.
And even on a V-Max the newer FAST version will bring in some major cash. You will find it as a separately billed option when you would like to use FAST for fat LUNs and thin LUNs.
However, one general statement was clear. "FAST is coming!"
There are no comments for this post