VMworld2015: RDSH Load Balancing


VMware Load balancing

At VMworld 2015 VMware announced an
improvement to RDSH load balancing. With the release of RDSH load balancing offered
was pretty simple not taking load of servers into account. There was no real
load being determined resulting in servers being offered session they should
never have gotten.
VMware now releases the second version of
their load balancing for RDSH and it’s improved quite a bit. They heard from
customers and learned from lost deals that you need to offer real load balancing
if you want to compete with Citrix.

Why Load balancing?

When you offer users a VDI desktop they get a
personal desktop with more or less dedicated resources. One user will not be
able to bring down other VDI desktops normally during the day.
When however you offer your users a RDSH
desktop, a shared desktop you face other challenges. One user on that RDSH
server could execute a program or task that would hinder others doing their
work.
That’s an unwanted situation and that’s why
next to managing what is started and what resources are used you need to manage
who connects to which server. Based on several metrics the load of a server is
determined. If that load is too high the server is not offered to users
requesting sessions.

Perfmon

VMware has created the option to use perfmon
scripts to determine the load of the servers. These perfmon scripts are
executed on the RDSH servers.
Per application you set the number of times
the application can be started. Once that number is reached the server is
removed from load balancing.

The power of Perfmon

At first I was like “why can’t they build a
normal GUI with some basic rules like Citrix does” but after a while I realized
that it’s a powerful option they’ve created. With creating your own perfmon
script you can combine both server and application metrics to determine the
load.

Return values

Scripts return values to tell you how they’ve
done, those values can be used in other scripts or management tools to show a
status of a component.
The perfmon scripts that will run on the RDSH
hosts will show a status ranging from 0 up to 3.
  • 0 is effectively a block of the
    server, saying please don’t use me in load balancing
  • 1 is showing that the RDSH host
    is not busy at all, 1 is the value for LOW
  •  2 is showing that the RDSH host
    is performing normally, it can handle more user session. 2 is the value for
    MEDIUM
  • 3 is showing that the RDSH host
    is having a hard time, the load is high. 3 is the value for HIGH.

My goal for this week is to find how exactly they will integrate this in the connection server for that is not clear right now.


Leave a Reply

https://tracking.cirrusinsight.com/869c29e2-3a9b-48c5-9232-0b95e7993ae8/controlup-com-pixel-php