May 13

EMC World 2010 is coming to an end. I am in my last session here and finally have some time to update you all on what I have seen. Really it is what everyone has seen as it was announced here. The star of the show, VPlex. This allows for active/active storage using synchronous distances.

The theme this year is “The Journey to the Private Cloud”. My company started this journey years ago so it’s nice to get a conference dedicated to what we have been working to accomplish for years. We have been looking for some technology to allow us to “flip the switch”. We want our workloads to be mobile and unbolted from the physical machines. This process started with VMware many years ago. As we close in on 95% of the workloads virtualized the next task is to unbolt the workload from the data center.

Now moving a workload from a data center has been possible for a while. However moving that workload back has been tricky. The main factor is time and effort. You could restore a work load and then update clients to point towards it. Then you had to do the same to go back. VMware SRM made this a little easier but it isn’t quite “flip the switch”.

Flipping the switch is an action. We want this to be a repeatable action. We want the effort reduce to a minimum as well as the time. Days and Weeks is not good enough for us to be able to move a workload. Hours is better but still not what we are looking for. We want minutes and seconds. We also do not want the workload to be halted when we “flip the switch”. This meant a few things had to exist that just were not possible until recently. You needed uninterrupted network access to the workload. You also needed writable access to the data in all locations. You then need the workload to move between physical computing devices without interruption.

Moving a workload between physical computing devices is possible with vMotion when using VMware. This has been in existence for many years but only inside of a data center.

Cisco has solved the issue of uninterrupted network access to the workload with Nexus. A new paid for feature for Nexus 7K’s called Overlay Transport Virtualization. OTV allows for stretching a vlan but stops all of the bad stuff your network people will tell you about doing this. The IP of the vm now does not have to change even though it is running in a data center that is 100km away from where it was.

EMC has solved the uninterrupted access to storage, both reading and writing. Active/Active access to storage with disks that are separated by 100km. VPlex allows for synchronous access to both storage arrays while presenting a single LUN to the vSphere host.

You then simply vMotion the VM and no one is the wiser. This is all brand new technology and information is trickling out about it. The next year or so will be very interesting with this new capability. EMC has said they are working on increasing to asynchronous distances which will open up a new world to mobile workloads.

Very interesting indeed.

May 13

EMC has a long track record of providing excellent tools to manage their products. Navicli is an example of a product that is included with your Clariion. EMC maintains up to date versions for any platform you can imagine, aix, linux, hpux, windows, vmware, etc. This tool allows you to perform many commands on your Clariion platform. EMC released some new technology called F.A.S.T which stands for Fully Automated Storage Tiering. This is really a bundle of 3 tools and is called F.A.S.T. V1.

The 3 tools are Navi Analyzer, Navisphere QoS, and Fast LUN Migrator. Let’s think about this for a bit. Everyone should already have Navi Analyzer. I always purchase it for my arrays. I also already own QoS. Now I am forced to buy them again to get the new tool Fast LUN Migrator. The reason for this? FLM requires .nar files to work! It cannot function with .naz files which you can always generate without Navi Analyzer. Rather than release the ability for you to analyzer .naz files with their tool and restrict you from actually opening them with an analyzer interface they force you to buy navi analyzer. I have yet to see why you actually need QoS as it is not used by Analyzer or FLM. It appears this is just an extra. The price for the suite you ask? You could buy a giant nexus one phone instead.

The installation of FLM is only 64MB, almost 1$ per byte. None of this is adding up to value for me yet. Let’s look at what FLM actually does for us. We already know what Analyzer and QoS brings to the table, the mystery is this new tool that we have to pay so much for.

FLM is a command line tool for windows. Here is how it works.
1. You download a nar file from your array.
2. You run fastlunmigrator.exe lunanalyze narfile.nar
3. You open a csv file and type in destination raid groups for the luns that are recommended to move.
4. You run fastlunmigrator.exe lunassist -h ipaddr -user adminuser -password adminpass -scope 0 -file narfileout.csv
5. You wait for the lun to migrate to the destination raid group. You can check status with fastlunmigrator.exe -status -h ipaddr -user adminuser -password adminpass -scope 0
6. Once it says complete for step 5 above, you are done.

Now what actually happened in step 1. Well you used analyzer to get the nar file if you did not own analyzer you would get a naz file. In step 2 the FLM product looked for luns with over 70% utilization or an average busy queue length of greater than 5 or a response time over 30 ms. If it finds one then it reports in the csv that it needs to move to faster disk type. In step 3 you MANUALLY choose raid groups and type them into the csv so that FLM knows where to move the source lun to. In step 4 FLM is creating a new lun of the proper size in the destination raid group, this is the destination lun. It then does a lun migration to this lun from the source lun. Step 5 is just a textual report of migration summary.

Now all of this can simply be scripted. Many shops may already even have this scripted. This Fast LUN Migrator tool should be free. The real power is analyzer. It is the one that is providing all of the data. Everything else can be done in Clariion manually. In fact you would likely do lun migrations one at a time anyways. Could you imagine if NaviCLI was a paid for product?

Now EMC has announced that F.A.S.T. v2 is coming out. This is sub lun tiering and sounds very interesting. There are some articles on the internet that mention EMC may be behind the curve here since some of their competition already have very similar features. F.A.S.T v2 and Flash Cache will be two emerging paid for features for Clariion to keep an eye on. Oh did I mention F.A.S.T. v2 is not an upgrade to F.A.S.T. in the sense you get it with your maintenance. No, F.A.S.T. v2 is a PAID for in addition to F.A.S.T. v1. They are going to hit you twice, that’s right folks.

EMC,
You have made a mistake here in charging for Fast LUN Migrator. This product should be free. I wasn’t overly surprised you charged for it at first because I thought it did more. At first I thought it would be sub $10,000. To find it is almost 6 times that list price is shocking. I keep looking for what you think is worth so much and I do plan to investigate more. Charge for F.A.S.T. v2 not v1.

May 3


I have been working on implementing this solution from Cisco. At first I was a little skeptical that Cisco made a good choice entering the server world. Cisco has done an excellent job in the network world. This appears to have carried over to the server world.

I’m curiously considering referring to Cisco as a data center company not a networking company.

I really do like the VCE combination. I’ve been running VMware and EMC in my data center for 5 years. After spending many weeks researching Cisco UCS and their competition I’m happy to add Cisco to the mix. So far everything they have designed into the solution makes perfect sense. The palo adapter has been the solution to many problems I have white boarded. The further I get into UCS the more I like it.