We are trying to benchmark ATS performance for our use. WE find that whenever we fetch 1M and 5M objects (with non-cacheable response) which do not get cached increase the latency so much:
What is the flow for big objects that are not cached and why the latency for them is much bigger than for typical objects(17k/15k)? What are all the parameters that I can play with, to increase the latency? Thanks & Regards Saraswathi Venkataraman | Xoriant Solutions Pvt. Ltd. Winchester, Hiranandani Business Park, Powai, Mumbai 400076, INDIA. Tel: +91 22 30511000 | Ext: 1113 | http://www.xoriant.com -----Original Message----- From: James Peach [mailto:[email protected]] Sent: Wednesday, July 04, 2012 11:07 PM To: [email protected] Subject: Re: ATS on worker mode On 03/07/2012, at 9:42 PM, Saraswathi Venkataraman wrote: > Is there any way we can compile the traffic server to use the MPM worker as > we do with the Apache webserver, to improve the performance? No, Traffic Server has a multi-threaded, event-driven work model. What performance issue are you experiencing that you expect a MPM model to help with? > > Thanks & Regards > Saraswathi Venkataraman | Xoriant Solutions Pvt. Ltd. > Winchester, Hiranandani Business Park, Powai, Mumbai 400076, INDIA. > Tel: +91 22 30511000 | Ext: 1113 | http://www.xoriant.com >
