High CPU utilisation on CAS servers in NLB on ESX host.

Hi. We had an ESX host that was hosting a number of vm’s which ran out of resources.  This ESX host hosted both our CAS servers. As this ESX host was the highest spec machine the decision was taken to keep both CAS server on this ESX host and move other vm’s. The performance of the remaining vm’s went back to normal but the RPC CPU utilization remained high until each CAS server was rebooted. Does anyone know why this would be the case? 
Additionally I am seeing repeated entries like this in the RPC Client Access log:
2012-02-16T11:49:14.587Z,177512,50858,/o=***/ou=First Administrative Group/cn=Recipients/cn=******,,OUTLOOK.EXE,14.0.6109.5000,Classic,,,ncacn_ip_tcp,,PublicLogon,1144 (rop::WrongServer),00:00:00.0156240,"Logon: Public,  in database a5f792b2-2897-4bc5-8063-3e6f142de9d9 last mounted on *******.******.****.local at 08/02/2012 22:16:03, currently Mounted; Redirected: not a user's home public server, suggested new server: /o=SHL/ou=Exchange Administrative Group (FYDIBOHF23SPDLT)/cn=Configuration/cn=Servers/cn=GMSMBX01",RopHandler: Logon:”
I have inherited this environment which is currently Exchange 2010 with one 2003 server. I am infering the 2003 box needs to be removed as it has no public folders. Also the 2010 public for the 2010 databases do reference the correct server, nothing is going over to the DR site.