To verify my previous statement, I also created log sink to BigQuery. With 12 000 rows in the table, please find the output of percentile queries below:
SELECT NTH(99, QUANTILES(httpRequest.latency, 101)) FROM [XXX:req_latency.appengine_googleapis_com_nginx_request_20180321] 0.99 latency: 0.023 seconds SELECT NTH(999, QUANTILES(httpRequest.latency, 1001)) FROM [XXX:req_latency.appengine_googleapis_com_nginx_request_20180321] 0.999 latency: 0.141 seconds SELECT NTH(9999, QUANTILES(httpRequest.latency, 10001)) FROM [XXX:req_latency.appengine_googleapis_com_nginx_request_20180321] 0.9999 latency: 0.715 There is definitely some big discrepancy here between what's reported by the AppEngine dashboard and actual logs latency. On Wednesday, March 21, 2018 at 10:03:48 AM UTC+1, Andrii Sydorchuk wrote: > > Hi Katayoon, > > Thank you for your message. I understand the difference between 95% and > log latency. > > The thing is that majority of logs have latency 20 ms, and there is only > 0.005% of logs with latency higher than 200ms. > > On Wednesday, March 21, 2018 at 1:06:30 AM UTC+1, Katayoon (Cloud Platform > Support) wrote: >> >> The 95th percentile and 99th percentile values in the dashboard tell you >> the point at which 95% and 99% of your traffic is experiencing latency >> while in the logs you see the latency for each request. >> > -- 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 https://groups.google.com/group/google-appengine. To view this discussion on the web visit https://groups.google.com/d/msgid/google-appengine/8e095fea-ad9a-44dc-b54f-96f48d5bcd03%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
