Archive for October, 2010


October 14th, 2010

I couldn’t have asked for a better night:  I attended the Minnesota Wild home opener with VMware, EMC, Tom Becchetti, Scot Joynt, met Paul Hokanson (TC with EMC), great customers, and the Wild defeated the Edmonton Oilers 4-2 in convincing fashion.  However, this was not the end of the evening coolness.  I checked my email when I got home and received the following notification from VMware:

Congratulations on passing the VMware Certified Advanced Professional vSphere4 Datacenter Administration exam!

I’m now a VCAP4-DCA.

On short notice, I was offered a chance  to sit the VCAP4-DCA BETA exam before it closed.  I drove 220 miles back in June to sit the exam.  I found the test to be extremely difficult and wasn’t expecting a passing score based on my experience.  I won’t go into the details now about the exam since I’ve already written about that previously.  Oddly, I sat the exam on 6/21/10, yet the date on the transcript shows 21-Jul-10.

I am pleased to have this exam in the books after previously thinking I would have to retake it.  It will allow me to focus on the VCAP4-DCD exam which will uplift my VCDX3 certification to VCDX4 certification.  Yes John Troyer, I am collecting them all.

Update 12/15/10: VMware has notified me that my transcript has been updated in the portal.  When I took a look, I saw I was awarded VCAPDCA-14.  I’m guessing this means #14.  If you don’t know what I’m referring to, VMware assigns sequential numbers to candidates who successfully meet the certification requirements, much like Microsoft did or still does (My MCP # from 1997 is 423097).  My VCP # is 2712 and my VCDX # is 34 (still not reflected in the portal).  On a podcast a few weeks ago, Jon Hall stated a new number would also be assigned for the VCAP4-DCD track.  I haven’t gotten the results of that BETA exam yet.

ESXi 4.x Installable HP Customized ISO Image DNA

October 12th, 2010

Those of you who are deploying ESXi in your environment probably know by now there are a few different flavors of the installable version you can deploy from:

  • ESXi 4.x Installable (the non-hardware-vendor-specific “vanilla” ESXi bits)
  • ESXi 4.x Installable Customized ISO Image (hardware-vendor-specific bits)
    • ESXi 4.x Installable HP Customized ISO Image
    • ESXi 4.x with IBM Customization
    • ESXi 4.x Installable Dell Customized ISO Image

Each of the major hardware manufacturers does things a little differently with respect to what and how they bake in their special components into ESXi.  There doesn’t seem to be much of a standard which the vendors are following.  The resulting .ISO file naming convention varies between vendors and even between builds from a specific vendor.  The lack of standards here can make managing a library of ESXi releases among a sea of datacenter hardware difficult to to keep track of.  It seems a bit careless if you ask me, but there are bigger fish to fry.

This short post focuses specifically the HP flavor of ESXi.  What’s the difference between ESXi 4.x Installable and ESXi 4.x Installable HP Customized ISO Image?  The answer is the HP ESXi Offline Bundle.  Essentially what this means is that if you install ESXi 4.x Installable, then install the HP ESXi Offline Bundle, the sum of what you end up with is identically equivalent to installing the ESXi 4.x Installable HP Customized ISO Image.

In mathematical terms…

SnagIt Capture

Where are these HP ESXi Offline Bundles?  You can grab them from HP’s web site.  Thus far, HP has been producing an updated version for each release of vSphere.  For reader convenience, I’ve linked a few of the most recent and relevant versions below:

In addition to the above, both ESX 4.1 and ESXi 4.1 on HP systems requires an add-on NMI Sourcing Driver which is discussed here and can be downloaded hereFailure to install this driver might result in silent data corruption. Isn’t that special.

vSphere Upgrade Path

October 11th, 2010

Old habits can be hard to break.  Such was the case today when I called out an individual for producing an ESXi 4.0 Update 2 upgrade document without referencing the requirement to upgrade vCenter 4.0 Update 1 to Update 2 first as a prerequisite. 

Up until the release of vSphere 4.0 Update 1 back in November of 2009, the VMware virtual infrastructure upgrade path was such that the vCenter Server was upgraded to the newer release, then the ESX(i) hosts were upgraded afterward.

As shown in the ESX and vCenter Server Compatibility matrix below, beginning with vSphere 4.0 Update 1, ESX(i) hosts can be upgraded ahead of their vCenter Server counterparts.  In fact, VMware allows a radically wider in versioning variation in that vCenter 4.0 (released May 2009, with no update) is compatible with ESX(i) 4.0 Update 2 which was released in June 2010, over a year later.

SnagIt Capture

After being corrected, I recalled hearing of this new compatibility some time back but the bits had fallen off the platter.  For the record, I’m not always right.  I’m fine with being wrong.  It happens plenty enough.  For me, it’s all about the learning.  Retaining the knowledge is an added benefit but isn’t always guaranteed if not used on a regular basis.

This mantra will provide some flexibility which may be needed to upgrade smaller groups of clusters or hosts (say for troubleshooting purposes) without impacting the centralized vCenter Server which in turn would impact the remaining clusters or hosts it manages by way of agent upgrades blasted out to each attached host.

Before you celebrate in the end zone Dallas Cowboys style, do note from the chart above that the upgrade to vSphere 4.1 reverts back to the old methodology of upgrading the vCenter Server first, and the attached ESX(i) hosts afterward.  In other words, ESX(i) 4.1 is ONLY compatible with vCenter Server 4.1.

Go Vikings!

Unisphere Client V1.0.0.12 Missing Federation

October 8th, 2010

A few weeks ago, the EMC Celerra NS-120 was upgraded to DART 6 and FLARE 30, in that order.  Before I get on with this post, let me just say that Unisphere is the bomb and offers at least a few opportunities for complimentary writing to give it the praise it truely deserves.  My hat is off to EMC, they answered the call (or was it the screams?) for unified management of unified storage. 

What was my opinion of the old sauce? 

  • Navisphere for CLARiiON block storage management was ok although it had a few bugs which forced a need to resort to NaviCLI once in a while.  Other than that, it looked old and was in need of a face/efficiency lift.  I’ve manged a few enterprise arrays from other vendors which have this same feel.  The biggest problem there being no end in sight of lackluster management or performance gathering tools.  Some vendors seem content with what they’ve always had which leads me to a few conclusions:
    • They don’t use their own software
    • The expectation is to use the CLI only
    • Hardware vendors can have outstanding hardware components but that doesn’t make them software developers
    • EMC has bumped it up a notch, at least with Unisphere – I can’t speak to Symmetrix management as I have no experience there
  • Celerra Manager for management of the Data Movers/iSCSI/NFS/CIFS was bug free, but very slow at times, particularly at first login.
  • Seasoned CLARiiON and Celerra TCs (as well as NetApp pros) might laugh at my tendancy to rely on GUI tools, but management of the storage is so few and far between, relearning CLI when a seldom task needs to be performed isn’t precious time well spent unless the tasks are going to be repeated often enough.

I’ve had some legacy Celerra software CDs sitting next to me in my den for several months (Navisphere, Celerra Network Server, etc.) and I will have no problem banishing them to the basement, probably not to be touched again until the next time the basement is cleaned out.  So look for some positive Unisphere posts from me in the future as I get the time.

Getting back on topic…  Earlier today I had finished taking a look at Nicholas Weaver’s SRM video.  Later, I was in the lab playing around with the EMC Celerra UBER VSA 3.2 (it’s the latest craze, you really must check it out).  I noticed a Unisphere feature Nicholas highlighted in his video which I don’t have on the Celerra NS-120’s build of Unisphere – the ability to federate storage array management in Unisphere via single pane of glass.

The Uber VSA has the ability to snap in multiple storage arrays into the Unisphere dashboard by way of an Add button:

SnagIt Capture

The Add button is missing in the Celerra NS-120’s build of Unisphere:

SnagIt Capture

The DART versions match at 6.0.36-4, however, the outstanding difference appears to be the Client Revision.  What’s worth pointing out is that the Add feature exists in the older client revision found in the Uber VSA, but is missing in the newer client revision found on the Celerra NS-120 which was upgraded a few weeks ago.

SnagIt Capture

I’m not sure if federation of multiple arrays was purposely removed by design or if it was an oversight, but it would be nice to get it back.  I should also point out that although federation appears to be missing for multiple arrays, it still exists and consolidates management intra of unified storage arrays consisting of CLARiiON block and the Celerra iSCSI/NFS/CIFS.

Update 3/4/11:  The Celerra NS-120 is now running DART 6.0.40-8, FLARE,7.30.10 (4.1), and Unisphere V1.0.0.14.  The Add feature to tie in multiple EMC storage frames into a single view is still missing.