hi marzia ok, can you confirm these new limits? previously, most people designed their apps to run in the 100-300ms range (datastore access included). when you started to go over 300ms, the warnings would start to show up in the logs. suffice it to say, it was reasonable to assume when no warnings showed up you were in the range that would scale very well on GAE.
are you now saying that the new average request target should be below 100ms (datastore access included)? if so, this seems to me like a significant reduction in GAE functionality/flexibility. am I reading this wrong? (entirely possible) thx brian On Feb 24, 4:55 pm, bFlood <bflood...@gmail.com> wrote: > well that seems like a change from previous posts. I thought people > generally quoted 200-300ms as a safe place to be for most of your > requests > > On Feb 24, 4:13 pm, Marzia Niccolai <ma...@google.com> wrote: > > > > > Hi, > > > 20ms is not considered CPU intensive, but once you get up in to the > > hundreds, it is. > > > -Marzia > > > On Tue, Feb 24, 2009 at 5:43 AM, bFlood <bflood...@gmail.com> wrote: > > > > thanks marzia > > > > I dont want to read too much into Nick's results above but is 30-200ms > > > now considered to be CPU intensive? > > > > cheers > > > brian > > > > On Feb 23, 4:20 pm, Marzia Niccolai <ma...@google.com> wrote: > > > > Hi, > > > > > This is done on a per-request basis. > > > > > -Marzia > > > > > On Mon, Feb 23, 2009 at 12:14 PM, bFlood <bflood...@gmail.com> wrote: > > > > > > hi marzia > > > > > > when this occurs, is the temp governor set for the entire app? or just > > > > > the handler that caused the high CPU warning? above, Nick said that > > > > > the typical request was 30-200ms so it seems odd that it would be > > > > > throttled > > > > > > I think this is better then the original high CPU reaction (throw > > > > > exception after limit) but if it affects *all* handlers it might turn > > > > > out to be far worse for apps where the majority of requests are well > > > > > below the high CPU threadhold > > > > > > cheers > > > > > brian > > > > > > On Feb 23, 2:54 pm, Marzia Niccolai <ma...@google.com> wrote: > > > > > > Hi, > > > > > > > Upon some further investigation, it seems that this is the result of > > > the > > > > > new > > > > > > handling of CPU intensive requests, more information about which can > > > be > > > > > > found here: > > > > >http://code.google.com/appengine/docs/quotas.html#Request_Limits > > > > > > > Specifically "Applications that are heavily cpu-bound, on the other > > > hand, > > > > > > may incur some additional latency in long-running requests in order > > > to > > > > > make > > > > > > room for other apps sharing the same servers. " > > > > > > > Essentially, if we observe that you have some heavily cpu-bound > > > requests, > > > > > > your handler may experience additional latency. This may not always > > > > > happen, > > > > > > and for the higher cpu request handlers, there is no way to know > > > exactly > > > > > > when it may happen. > > > > > > > -Marzia > > > > > > > On Wed, Feb 18, 2009 at 10:21 AM, Nick Winter <livel...@gmail.com> > > > > > wrote: > > > > > > > > We've been seeing pretty brutal response times intermittently for > > > our > > > > > > > app (id: skrit), which I can't figure out. Some more info: > > > > > > > > * They're clustered in time: all requests will be slow for several > > > > > > > seconds at a time, dozens of times over a couple hours > > > > > > > * The long periods of slowness often happen in the morning, > > > although > > > > > > > not always, and not every morning > > > > > > > * Traffic doesn't appear to be related, and is low (less than 5 > > > > > > > requests per second) > > > > > > > * It's not instance startup costs. I'm logging when things are > > > > > > > imported, and on a small request with no datastore interaction: > > > > > > > -- with no startup costs, normally around 30-200ms, spikes to > > > > > > > 1300-1600ms > > > > > > > -- with startup costs, normally around 1300-1600ms, spikes to > > > > > > > 4000-11500ms > > > > > > > * It doesn't seem to be related to dynamic get latency or anything > > > > > > > else on the serving status page. Initially, we thought it was > > > loosely > > > > > > > correlated, but I think those were just flukes. This happens much > > > more > > > > > > > often, and for much longer, than serving latency is ever high. > > > > > > > * It happens on all requests; I've just given numbers for > > > > > > > simplest/ > > > > > > > most common request. > > > > > > > > As far as I can tell, this has only been happening for the past > > > month > > > > > > > and a half or so. > > > > > > > > Any ideas? It's as if we're just intermittently being given really > > > > > > > slow serving.- Hide quoted text - > > > - Show quoted text -- Hide quoted text - > > - Show quoted text - --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Google App Engine" group. To post to this group, send email to google-appengine@googlegroups.com To unsubscribe from this group, send email to google-appengine+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en -~----------~----~----~----~------~----~------~--~---
