NTWS bogging down. How to best scale with SQL Server

[Mike Grigsby]

A client is using SQL Server Express with one of my programs. He's bogging down at around 1000 of his customers. He wants some ideas on how to scale the app better.

His clients put about 300mb of data and images on the server, on average, and they are consumers who load up the system in the evenings and weekends.

I think the problem is the processor handling the requests and uploading. Does anyone have any experience using the Data Center version of SQL Server, and would using more processors help? I think it will allow up to 256 processors, assuming that's the issue (I think it is. He has tons of bandwidth and disk space).

Conceptually, I'm not sure how to test this. Would he run multiple instances of my NTWS software? I've got some other ideas for scaling by re-routing and balancing among different computers, pipes, etc., but the client is hellbent on using Data Center, and I could use some direction.

Also, if anyone has substantial experience at Data Center, please send me your resume. Thanks!

[David]

Mike,

You might look at the memory utilization.  NT caches uploads into memory until the whole file is there and then writes to disk.  I found that my server stalled out trying to upload a single 800mb file.  If there are lots of people uploading smaller files, depending on timing, you could be running into a memory issue.  I ask Bruce if he could add streaming the file to disk onto his todo list.  I'm hoping it's moved closer to the top.

Here is the link to that thread... http://www.nettalkcentral.com/index.php?option=com_smf&Itemid=36&topic=1705.0
 

[kevin plummer]

Hi Mike,

There are some restrictions with SQL express in terms of performance.

You could do some testing with SQL Developer edition (which is pretty much the enterprise version but cheap to buy) which doesn't have restrictions on number of processors and if that solves the problem then upgrade to enterprise. Data Centre version I think is more for virtualization and installing multiple copies.