SQL Server Performance

Performance of Reporting Services

Discussion in 'SQL Server Reporting Services' started by hmckillop, Aug 23, 2004.

  1. hmckillop New Member

    Has anyone used Reporting Services outside of an Intranet Environment where performance was an issue. If so could you give me a few brief pointers.
    I have noticed that excessive app to DB calls happen when a single report is executed and although ok in an intranet environment not exactly ideal for a heavy load production web environment
    Thanks in Advance
  2. hmckillop New Member

    Ok not much of a response - so I decided to use our load tool and test this for myself with the following results.

    I have 2 environments, ENVA and ENVB
    ENVA - is a 3 tier setup, with Web (ASPX and C#), App(C#) and DB(SQL 2000 - 4* Pentium4 2.8gHz) on seperate servers (all Windows 2000 Prof).
    ENVB - Reporting Services Server, DB server both SQL 2000(both 4 * Pentium4 2.8gHz) on Windows 2000 Prof.

    I created a simple report - i.e. a single parameter passed from web down to a stored proc on a db via an App layer and then return the resultset and display,
    This report was created on the ENVa and then a simple report (created using the Visual Studio 2003 wizard) to call the same stored proc with the same parameter.
    Both of these were executed using QALoad under 50 virtual users (stepped increase up to max of 50, 1 new user every 5 seconds) and a sleep time of 5% (essentially means I am replicating 1000 concurrent users). I also run the same test only for 1 user to give a baseline.

    Under ENVA - there was little degradation of performance as the new users came on to the system. (CPU didnt go above 30% on any of the boxes)
    On ENVB (i.e. the reporting services setup) - a significant degradation occurred at around 25 virtual users, with large failure rates happening at around 35 virtual users. (Near 100% CPU usage on all CPUs on Reporting Services Machine)

    After a bit of research into the differences, essentially Reporting services calls about 10-12 stored procedures for execution of one actual report. This saps time and resources, managing seperate RPC calls to the Reporting Services database. I can post more detailed analysis findings if required - but need to look into the caching abilities of Reporting Services.


  3. backborn New Member

    Hi,
    I have met the same problem,It took me 2 hous to solve this problem.My friend
    recommend me to use RAQ Report.RQA Report is so excellent that I love it very much.
    After I use RQA Report in three months ,I find that RQA Report is better than Crystal Report in many aspects,such as
    it has a faster speed and a smaller CPU utilization.[FONT=宋体]You can get it at [/FONT][FONT=宋体]http://www.raqsoft.com/[/FONT][FONT=宋体] freely.[/FONT]

Share This Page