Don't Fear The Recompose!

Hi Friends,

I needed to update my golden image in VMware Horizon View 6, so that means a Recompose.  For those not familiar with the term, Linked Clones have a very clever method to keep the size of your desktop to a minimum.  Only the changes are recorded to a delta disk which can grow over time.  To optimize their size, you run a Refresh.  Now say you update your golden image, like I just did, you then run a Recompose which creates new master clones and updates all of the Linked Clones.

I wanted to see how my Nimble CS400 series would handle the Recompose of 400 desktops, since the process is doing quite a bit of work.

The graphs are very encouraging, especially the latency and cache hit ratio.  The latency is low and the cache hit is almost 100%!  What does that mean?  Desktops getting recomposed super fast my friends!



While watching the Nimble GUI I was also watching the vCenter GUI and the Recent Tasks were just flying by.  So I decided to record about 50 seconds of tasks and you can see the processes just completing almost as fast as they are started.  The timing on the video was NOT modified, this is real time.

The Nimble is just plowing through the Recompose and not even sweating.  :-)

(Image borrowed from Google)

Kind of like how this Ford Raptor is plowing through and flying over the landscape!

Until Next Time!
-Brain

Darn You Networking! You've Foiled Me Once Again! (Emotions Expressed by Homer Simpson)

Hi Friends,

First and foremost, thank you to the great Matt Groening for bringing us The Simpsons!  All images borrowed from Google.

From my title you can see I'm having a moment with my network.  Not the switching our routing, but general "networking" concepts.  The things so simple we, or at least *I*, overlook....

If you've been following my blog you know I've been running performance numbers in my VMware Horizon View 6 environment with Login VSI.  I got the environment working well with 100, but it was time to increase the number of desktops.  It's always a challenge when you start increasing the number, not because the products are bad or anything, but because of all the darn little moving parts!!!






In a testing environment you're deleting, creating, re-creating, deleting, etc. the zombies start coming out everywhere!  So every so often you've got to go medieval on their butts, get out the fire hose and just pressure wash the environment.








So after one of these power washes I re-created all of my desktops, upped the count to 400 and kicked off a Login VSI run.  The Basephase started and 2 of the 5 sessions kicked off.   Hmmm, weird....  I rebooted all of the desktops and the launchers, and still, 2 of the 5 kicked off...  Think Neil, THINK!


I took a look at the desktops that were running and sure enough, 2 were happily running the workload, but 3 were just sitting there at the Windows 7 desktop...

I started searching Google and found some cases on the Login VSI site that were similar to mine, but the customer's couldn't get anything to launch.  My problem was I could get some of them to launch.  Lousy rotten intermittent problems!!


Hmmm, what could it be?!?!  I logged onto one of the machines that didn't kick off the workload and did and nslookup of my share machine.  Yep, that works....  I then did a ping and no ping....  Then it hit me!  My DHCP environment is larger then a Class C BUT my share machine is set to 255.255.255.0.  I changed my netmask to a Class B or 255.255.0.0 and my client machines outside of the Class C could now ping the share machine!!




So sometimes the smallest things can REALLY throw a wrench in the gears.

Until Next Time!
-Brain