Once ENT Server has received an audit snapshot, it then needs to process this information and/or store it. Under the hood, Enterprise Server makes use of either an MSDE or MSSQL Server database server to store and retrieve information about the IT Assets that you manage.
The first step in this analysis is to check if the machine already exists in your IT asset inventory. If it does, then ENT Server retrieves the details of the existing PC from your Inventory and compares these to the information contained in the audit snapshot that it's just received. It then sends out any alerts for faults, changes or new and missing hardware components and software. If the PC doesn't already exist in your IT Asset inventory then ENT Server will simply add a new PC from the audit snapshot.
The analysis/storage/retrieval process described above is performed by issuing a series of SQL statements using the ADO database engine. Now, compared with WMI (to which we issue WQL statements during a PC Audit), MSSQL Server is a high performance beast, so the times involved are typically much less. However, issuing SQL statements and receiving the results of these may still involve the transfer of a fairly considerable amount of information, so it certainly pays to have your Enterprise Server on the same LAN network or even the same machine as your MSDE or MSSQL Server database server.
Additionally, if only one Enterprise Server and only one SQL Server are responsible for processing all of the audit snapshots collected for all of the computers in your organization, processor power and/or memory capacity on the machine(s) where your Enterprise Server and MSSQL server are installed may also have a notable impact on the performance of the overall auditing solution.
You'd really have to monitor the processor/memory usage on the particular machines where your Enterprise Server and SQL Server are installed in order to draw any conclusions about whether it would be necessary to scale these particular components to 2 (or more) machines. On a network with 10s of thousands of machines, scaling of these server components to multiple machines, as well as making provision for fault tolerance and redundancy, may certainly be worth considering.
