Can you suggest the best way to create VS for a single API with 1000 Request Response pair with minimum performance degradation?
I don't know if this is the 'best way', but if you were able to put all 1,000 request/response pairs in a single folder, then you could use the 'Create VSI using request/response pairs', and let LISA create the image for you. LISA would be able to identify likenesses between the request/response pairs and choose where magic strings makes sense. After the recording was done, you may only end up with 20, or 10, or even 1 request/response pair that works for all 1,000. However, I am not sure what you mean by minimum performance degradation. If you mean that you want to keep the performance of your VS identical to live production performance there are a few ways to accomplish this. One is to use the VS companion that plugs into Wily (if you have Wily). This will allow you to pull down response times for any given block of time from your production version of the API. Then, when you run the virtual service it will follow the average response times that are provided from the block of time you specified. Another way, is to us the VS companion, but instead of plugging in Wily, you can manually create a response time pattern for the VS to follow. You can try to mirror that as close to production as possible. The final way to accomplish this is to manually go into each request/response pair in your VS, and set the think time to exactly what you want.
Many thanks for your reply. Just to make problem statement more clear -
We have distinct 1000 request and response pairs. Which means my operation in VSI will mostly have 1000 different data combinations in exact match. Once I deploy this sort of VS (which I feel will be quite bulky), we are worried about performance of Virtual service (response time should be quite low as this is used for performance testing). I am afraid that due to huge no of data scenario, LISA may take time to find exact combination and hence response time may go bit high (from normal 10 ms to say 100 ms). Will that be the case or LISA can handle any no of data without any issue on performance? If there is possibility of performance degradation, can we have some solution to mitigate this please?
My VSE is created for performance activity in line to CA recommedation only.
I believe that the answer to your question will be found in how big your VSE is. Of course, if your VSE is small in size (both RAM and CPU), the service may suffer to the point where your VSE hangs because of insufficient memory. My VSE’s are physical Linux boxes with 12 core CPU’s and 98GB RAM. So, I would likely not see any performance degradation with a service that holds 1000 xml request/response pairs. Since I don’t know the size of your environment or how much an environments degrades based on the size of a VSM, I would look to CA Support to give you some guidance on what kind of performance to expect when using a VSI with 1000 request/response pairs.
Hope that helps!
Jeff De Meyer
Retrieving data ...