We have a virtual environment using Vmware. All servers are running on a windows 2008 R2. The virtual environment sits on a HP Proliant GL380 G7 server, with 1 x Intel Xeon E5640 / 2.66 GHz. There are 8 disks on there. The first three are SAS 10,000 disks. The other 5 are SATA 7.2 disks.
Vm1 which is the actual remote desktop server, has been installed on the SAS disks. On this server, MSoffice is installed. It has four cores allocated to it. 8gb of RAM has been allocated here.
Vm2 is the file server installed on the Sata 7.2K disks. This is approximately 2TB in size. It stores lots of picture files and imagery, along with excel files. 4Gb has been allocated to it.
The file server is mapped to other users and used by the entire company totalling around 220 users daily. We have approximately 50u sers logging onto the remote desktop server (Vm1) per day and accessing the files on the file server (Vm2). Once logged on, the majority of users begin to use Outlook, Word, Excel and print from this server also.
We've got a dedicated 4mb up/4mb down leased line associated with this server. I've checked and the majority of the time, the CPU load is below %40 and RAM is at around 6.4gb from the allocated 8gb. Whenever people start using it, they do complain about the actual 'slowness' of the remote desktop. There is sometimes a delay when people write emails from the time they type and it shows on the screen or just opening up of excel files.
1) Can it be that there is a lacking of disk IOPS due to the number of people accessing these servers?
2) If you had the choice based on the approximate usage, number of people logging on, how would you configure your servers? Is there anything you would do on the core switches they are connected to? (like link aggregation or ensuring there is a 2gb/s output on the switch)
Any ideas that you could throw in there, would be great. Remember, I'm not necessarily trying to troubleshoot but know mainly how you guys would go about setting your environment up.