Optimise AdminKit for large network, 3500 clients on adminkit Options

[mlpotgieter]

We have Adminkit running on Windows 2003 server with MSSQL2005 (full version not express). The server is 2xQuad core with 4 gigs RAM and 1 Terabyte drive space in RAID arrays with SAS drives. Also have about 1000 groups, and will have 3500 clients (workstation and enterprise server) when the rollout is finished. We are also using update agents and have about 500 of them setup on the groups. We are only using IP subnetwork scanning and have setup about 500 IP subnetworks which get imported into the relavent groups.

At times the Adminkit becomes unresponsive or VERY slow or we can not even connect via the MMC snapin, even if the CPU and Memory utilisation on the server is very low.

Is there anything that can be done with Windows2003, MSSQL2005 or Adminkit to improve performance?

We have only rolled out 700 Workstation and Enterprise Server clients (2200 network agents) and performance is not good. I have logged a case with Kaspersky Technical support on April 21st but support have not given me a suitable answer, they only said I must use slave servers even though in the same response they say the number of groups the Adminkit can handle is unlimited (resouce dependant)..

So can the Adminkit not handle 3500 clients divided into 1000 groups? Do others on the forum have large, distributed networks setup?

[Tybilly]

Hello,

Do you have a policy in each of your logical groups ? Try to reduce to number of logical groups, and the number of policies / tasks.
I had the same issue with a customer, and the only solution is to have slave administration servers. The problem is not the number of managed hosts, this is the number of logical groups, policies and tasks which take a lot of resource.

Or you can wait for the next version 8.0 of Administration Kit which seems to be optimized for this kind of structure (end of Q3 2009).

[mlpotgieter]

Do you have a policy in each of your logical groups ?

We only have one set of policies at the top of the heirarchy. The only reason we have groups is for update agents. I didn't want to install 500 slave servers but it seems as though this is the only way this will work.

I would like to understand where the bottle neck is... the server is not using even half of its resources... I am begining to think that it is an issue with the Adminkit speaking to either SQL or the network.