Hope you used bind variables. If not you won't get anywhere close to 1,000
concurrent users.
Doing this type of load testing is difficult and time-consuming. There are
tools out there that will simulate the load, but often they do not mimic the
way you are actually issuing the statements. You can get a tool that will
drive your application and thus the mimic is the same, since it is your
application. You need to do (from a high level):
1. Come up with a usage or workflow of your application. That is how are
people going to use this application. Model the reasonable ways they are
going to use the application. In your case probably ordering, searching for
order statuses, etc.
2. Find a tool that will drive your application e.g. Silktest. Then spend a
lot of time setting that up and putting in random parts.
3. Get or rent enough hardware to simulate the test.
4. Spend a lot of time doing 1,2,3 and doing the simulations.
Jim
"tshen" <tshen01_at_hotmail.com> wrote in message
news:5Pd78.66977$B94.13228040_at_news02.optonline.net...
> hi, Guys:
>
> We have just built a customer ordering application which use oracle as
> backend DB system, our goal is allow 1000 concurrent connections and
> transactions, here are the questions, I want ask your experts:
>
> 1) how can we do the simulatation and benchmark for 1000 concurrent
> connections?
> is there any tools we can use?
>
> 2) How can we tell the current hardware can handle the 1000 concurrent
> connections?
> if it cann't then what we need more?
>
> Thanks.
>
>
Received on Sun Feb 03 2002 - 11:18:05 CST
