Precise/Indepth for SQL Server
Last month I passed on the initial briefing information on Precise/Indepth
for SQL Server. This month I had a chance to take the software for a spin.
As you'll recall, this is a new performance monitoring and optimization
solution for SQL Server. Precise worked with Microsoft to hook up a new
high-speed, low-impact sampling technology to the SQL Server engine. This
allows the new product to collect and analyze very detailed performance
information in the short term and over time, to identify both immediate
problems and long-term issues. The Precise screen can show you which statements
are being executed, which ones take the longest to execute, and how tables
and other objects are bring used. You can summarize resource consumption
by program, user, or type of application for a business-centric rollup
Installation of the software is fairly complex. There are databases to
create, services to install, and a client piece to see what's going on.
You also need to install a SQL Server hotfix with the patch from Microsoft.
This isn't for the faint of heart (it requires replacing sqlserver.exe
and running a script to change some system objects), but it worked flawlessly
on my test server. Once everything is in place, you can control the server
agents from the client, and of course everything is stored in a SQL Server
"Everything" is quite a lot. You get statistics on every statement that
the server executes, together with graphical information on which resources
the statements are waiting on. All of this information is stored in a
warehouse database and aggregated to make it easy to spot trends or to
determine what was going on at a particular time. All of the information
is sliced very finely, so it's easy to track problems to a particular
user, query, application -- or even to a particular step in a query plan
taking excess locks. The manual is thorough and contains valuable hints
on performance troubleshooting, but I was able to point and click my way
through what was going on in my test server with no trouble.
With an entry level around $3m000, the target for this product is the
Fortune 2000. By making it simple to move from aggregate data to detailed
statistics, this will be a welcome addition for any DBA trying to troubleshoot
performance issues on a heavily-loaded server.
[This review originally appeared in developer central 1.11.editor]
Mike Gunderloy, MCSE, MCSD, MCDBA, is a former MCP columnist and the author of numerous development books.