The other the Communicator Web Access web site application kept failing.  If the application was started, the moment someone tried to access the site the application was disabled by IIS.

The error that was generated was  "The identity of application pool, 'CWAApplicationPool' is invalid and error code 80070532"

The solution was very simple.  The service account did not have the password never expires option set.  All I did was set this AD account option, and the application pool stopped being disabled.

Recently I wanted to have a play with Data Protection Manager (DPM) to look at its features for a hosted demo environment, but didn't have much of a budget.  I had a server with little disk storage space available, and no real option to add any more drives.

Without much thought and without doing much research I went out and bought a WD 1TB USB external drive.  However, on installing DPM I noted that it did not see the drive.  Upon 5 minutes reading the help file, DPM doesn't support USB or Firewire drives.

Wasted money?  Not quite.  I simply ran up HyperV and installed DPM into a virtual machine.  I then created another vhd file on the WD external drive and mounted that on the HyperV VM running DPM - which recognised the mounted vhd file based drive as valid for DPM.  Voila!  Access to storage to back stuff up.  May not be fast, but my it achieves my objective of a cheap informal backup solution.

If you are seeing event log errors that looks something like this on your MPS engine server:

"Resolution of execute statements failed from a procedure=DeleteOrganization_ namespace=Managed Helpers executing procedure=TryDeleteAddressLists_ namespace=Managed Helpers. Check Event Viewer for additional namespace load errors."

It appears this is the result of some obsolete procedures calling non-existant ones.  Check out this forum thread for info on how to rid your event log of these errors : http://forums.asp.net/t/1137212.aspx

Happy hosting!

Ben

Here is a little tidbit you might not be aware of.  MPS doesn't create a GAL or OAB for an organisation until you provision the first MAPI or Outlook Anywhere (RPCoverHTTP) enabled mailbox for that organisation.  This means you don't create it unless you need it - which means you don't waste a GAL and OAB for organisations that only have POP or OWA mailboxes.

You could say it is just implemented common sense really.

Happy hosting.

Ben

Having seen and done a lot with Windows Network Load Balancing(NLB), I thought it might be helpful for those who aren't familiar with NLB if I wrote a series of posts which explain what it is and how it works - from the beginning.  To start the journey, this post will explain, at a basic level, how it works and why it is different to other solutions.  Future posts will explore more detail on how it works and various ways you might use it for your hosting solution.

So what is it?

Simply put, it is an algorithm that a group of servers use to determine which server will process inbound network traffic at a point in time.  English?  Well, perhaps I'll explain what it is not first.

You may be forgiven for thinking it works similarly to a hardware load balancer - a device that logically sits in front of the servers you are hosting, and then directs each incoming request to a server according to rules set on the device.  The hardware load balancer will ensure that the traffic load is distributed across the servers.  Or to put it another way, it is a traffic cop that directs each network packet to a particular server so that each server is receiving an equal number of requests.  This is an overly simple explanation for illustration only.

NLB does not do this - it uses a different strategy to achieve the same goal.  When you use NLB, every packet is delivered to every server.  What NLB then does is use an algorithm to decide which server will process a particular packet, and the rest of the servers will drop the request and not process it. 

So, in summary, a hardware load balancer directs individual packets to a certain server in a set of servers, whereas NLB works by all servers receiving all packets with only one server processing the packet. 

How does NLB achieve this?  Well, this will be the focus of the next blog entry on this topic.

Happy hosting.

Ben