On Saturday, September 14, 2013 2:57:35 PM UTC+8, aswath wrote: > > If a request a served by an instance that is alive for sometime, then it > is fine. > However, sometimes the request is directed to a new instance. Any first > request for a new instance is adding a considerable delay. > > This implies you have some control over the performance. As I said earlier either it's something that is expensive to initialise and is not occurring as part of the warmup request or you are caching something expensive as a result of the first request serviced by the instance.
I would strongly recommend you us appstats and logging, and determine where in your request processing the time is consumed on this initial request - maybe you can get whatever it does performed as part of a warmup request. T > -Aswath > > > On Sat, Sep 14, 2013 at 10:50 AM, timh <[email protected] <javascript:>>wrote: > >> I would check that this is always the case. Also check what 3rd and 4th >> request is like - make sure it's the same instance. >> >> If so it would then suggest that additional initialisation is being >> performed that is expensive or that caching of some data (memcache for >> instance) has a significant performance impact (improvement) for 2nd and >> subsequent requests. >> >> You should use appstats to profile what this request is doing. >> >> T >> >> >> On Saturday, September 14, 2013 1:05:49 PM UTC+8, aswath wrote: >> >>> I am attaching the new logs. >>> >>> I was examining the logs. It did not say, it started a new instance. >>> But, there was a instance startup before that log >>> /_ah/warmup 200 15380ms 0kb >>> >>> I noticed that the request that is taking longer was served by the >>> newly started instance. >>> >>> >>> -Aswath >>> >>> >>> >>> On Sat, Sep 14, 2013 at 9:48 AM, timh <[email protected]> wrote: >>> >>>> You truncated the log for the 4159ms request. Did it say it started a >>>> new instance ? >>>> >>>> T >>>> >>>> >>>> On Saturday, September 14, 2013 11:15:37 AM UTC+8, aswath wrote: >>>>> >>>>> Hello, >>>>> I am finding a huge time difference for serving the same request. >>>>> Attached is the logs that shows the serving time for the same request. >>>>> The same request was accessed with in 1 min interval from the same >>>>> browser. There is no business logic change that happens between the >>>>> first >>>>> 1st request and the 2nd request. >>>>> >>>>> For the first request the serving the time is 4159ms, and for the >>>>> second request it is 832ms. Why is this huge time difference? >>>>> >>>>> -Aswath >>>>> >>>>> >>> > -- You received this message because you are subscribed to the Google Groups "Google App Engine" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/google-appengine. For more options, visit https://groups.google.com/groups/opt_out.
