Dear All,
I want to know how many CPU resources are required based on the number of http requests (or cloud workloads). For example, suppose at time 't', 'n' number of requests have arrived. Is there any methodology to predict the number of CPU resources for that 'n' requests?
I will be highly grateful, if anyone provide me the answer.
I'm not sure if there's an simple rule of thumb since it heaviliy depends on the implementation of the http server and the amount of data which is requested. But I'd not expect the CPU rather than file i/o or network will be the limiting factor.
Likely you will find some benchmarkes looking for specific implementations like e.g. the apache2 http server.
Hi Tanvir
Do remember that all CPUs are not equal. Also bear in mind that in the process of virtualising a CPU, there will be an element of overhead to take into account. A virtualised CPU will always be slower than the same CPU running in a non-cloud environment. And a more powerful CPU will always run faster than a less powerful one. I have attached a list of some papers on AWS performance. Of these no 1 on the list, Ou et al, in 2012 carried out an analysis of AWS cloud instances and found a large variance between resource performance. Here is the reference:
Z. Ou, H. Zhuang, J. K. Nurminen, A. Yla-Jaaski, and P. Hui, “Exploiting Hardware Heterogeneity within the Same Instance Type of Amazon EC2,” in Proceedings of the 4th USENIX conference on Hot Topics in Cloud Computing, 2012, pp. 4–8.
That should give you some useful background.
Regards
Bob
enc
- Badges
- Science method
- Similar topics
- Methodology