SPDY over HTTP doesn't give desired results every time. Need to know if testing is proper

[Sujan Narvekar]

Hello,

We have been evaluating SPDY for the last few days. However, we are not seeing consistent improvements of SPDY over HTTP as desired. We would like to know if our setup & testing is proper or do we need to do something else to get the desired results.

Appreciate if someone can help on this.

Details of the Setup :

1. HTTPS - 

F5 LoadBalaner (where SSL config is done) -> Apache Webserver

2. SPDY - 

F5 LoadBalancer (where SSL & SPDY config is done) -> Apache Webserver

Details of Testing :

Page 1 : Static page with Large no of Secondary Requests (Only Images) Page Size : 2.3 MB, No of Requests : 72

Page 2 : Static page with Large no of SMALLSIZED secondary requests Page Size : 86.1 KB, No of Requests : 71

Page 3 : Static page with Single but HEAVY secondary request Page Size : 12.2 MB, No of Requests : 3

Page 4 : Static page with Large no of secondary requests (CSS,JS & IMGS) Page Size : 3.5 MB, No of Requests : 72

We have also tried implementing mod_spdy on the Apache, however our current architecture has SSL done on the Loadbalancer layer. Hence we would like to implement SPDY on the LB along with the SSL.

We would also like to know what are the page sizes / no of requests / network utilizations considered when standard benchmarking of SPDY was carried out.

Regards,

Sujan Narvekar