If it scaled linearly like that, we probably wouldn't have problems with long running requests. Unfortunately, long running requests are bad for the ecosystem because they impose a non-linear cost.
The number is officially 1000ms. We have been saying 800ms because we allow for some variance. If you tuned your requests to be 990ms and had a period of 10ms of latency, you'd be dead in the water. 800ms is a safe enough number that even if you experienced an additional spike of 100ms-150ms for whatever reason (datastore slowness, unusual usage patterns in your application causing Memcache misses, network latency via URLFetch), you can tolerate it and be fairly confident you will be autoscaled. On Wed, Sep 15, 2010 at 6:51 PM, Flips <[email protected]> wrote: > @Harshal > Actually slower requests mostly consume more cpu time and are much > more expensive by default.. > > On Sep 15, 8:28 pm, Harshal <[email protected]> wrote: > > I am OK with Google introducing tiered pricing for handle this issue. > Don't > > take these numbers at their face values, but you would get the point I am > > trying to make here. > > > > Avg. Requests CPU Charges > > > > < 700ms $0.02/hr > > < 1500ms $0.04/hr > > < 2000ms $0.06/hr > > > > For all the requests Google provision new servers but if you requests > take > > longer you pay higher. Not sure if it really makes sense, but the idea of > > totally not allowing any scaling up is not good enough motivation to > write > > ever more complex apps. > > > > On Wed, Sep 15, 2010 at 10:51 PM, Jeff Schwartz <[email protected] > >wrote: > > > > > > > > > +1 and a whole lot more :( > > > > > While it is all our goals to produce efficient applications that can be > > > scaled out, the platform itself has to be usable &, might I add, > enforce > > > ceilings that don't choke the life out of even the simplest of > processes. In > > > that regard I'd be willing to give up a little bit of scalability for > > > somewhat more relaxed quotas. > > > > > But the real issue I believe is that of imposing unrealistic quotas. It > is > > > one thing to show an example of an efficient application built by > Google and > > > another to show how that relates to real world applications that though > they > > > employ all the same best practices still cannot function within the > > > allowable quotas. > > > > > Resiliency is also a major issue on App Engine, if 99% of our code is > > > protect the app from what can go wrong and that eats up our quota, what > is > > > left for doing real work? > > > > > It is my desire and I suppose that of many if not even most of the > other > > > developers that Google rethink their approach to providing scalability > & > > > resiliency to the masses on App Engine. > > > > > Jeff > > > > > On Wed, Sep 15, 2010 at 12:41 PM, Gordon <[email protected]> wrote: > > > > >> bothering, indeed.. > > > > >> On Sep 15, 6:11 pm, Robert Kluin <[email protected]> wrote: > > >> > I am starting to get concerned. A few months ago this number was > > >> > 1000ms, right? Then about a month or two ago it became 850ms; > > >> > actually I have even saw the 850 number posted within the last week. > > >> > Now it is 700ms? > > > > >> > From my experience, getting or putting even a single entity can use > a > > >> > substantial portion of 700ms (20% to 40%). If you operate on > multiple > > >> > entities you'll easily use 1/2 of 700ms. Just the act of _running_ > a > > >> > query takes around 250ms -- when the datastore is actually > functioning > > >> > correctly. > > > > >> > This trend is _really_ not good. > > > > >> > Robert > > > > >> > On Wed, Sep 15, 2010 at 10:18, bFlood <[email protected]> wrote: > > >> > > not for nothing, but isn't "we may not schedule additional servers > for > > >> > > your app" throttling? > > > > >> > > when did 700ms become a magic number? > > > > >> > > On Sep 15, 9:33 am, "Nick Johnson (Google)" < > [email protected]> > > >> > > wrote: > > >> > >> Hi, > > > > >> > >> We don't throttle apps. If your average latency is over 700 > > >> milliseconds for > > >> > >> user-facing requests, we may not schedule additional servers for > your > > >> app, > > >> > >> however. > > > > >> > >> What leads you to conclude that your app is being throttled? > > > > >> > >> -Nick Johnson > > > > >> > >> On Tue, Sep 14, 2010 at 9:47 PM, Coding Social < > > >> [email protected]>wrote: > > > > >> > >> > Hi, > > > > >> > >> > I have had appid mapthislink for many months now. Recently my > > >> > >> > extensions that use this web service to unwind urls have been > > >> featured > > >> > >> > by Google Chrome and Apple Safari so usage is up substantially. > > > > >> > >> > Can someone turn off the throttle? Causing latency and 13% > error > > >> > >> > rate. > > > > >> > >> > Thank you. > > > > >> > >> > -- > > >> > >> > You received this message because you are subscribed to the > Google > > >> Groups > > >> > >> > "Google App Engine" group. > > >> > >> > To post to this group, send email to > > >> [email protected]. > > >> > >> > To unsubscribe from this group, send email to > > >> > >> > [email protected]<google-appengine%[email protected]><google-appengine%2Bunsubscrib > [email protected]><google-appengine%2Bunsubscrib > > >> [email protected]> > > >> > >> > . > > >> > >> > For more options, visit this group at > > >> > >> >http://groups.google.com/group/google-appengine?hl=en. > > > > >> > >> -- > > >> > >> Nick Johnson, Developer Programs Engineer, App Engine Google > Ireland > > >> Ltd. :: > > >> > >> Registered in Dublin, Ireland, Registration Number: 368047 > > >> > >> Google Ireland Ltd. :: Registered in Dublin, Ireland, > Registration > > >> Number: > > >> > >> 368047 > > > > >> > > -- > > >> > > You received this message because you are subscribed to the Google > > >> Groups "Google App Engine" group. > > >> > > To post to this group, send email to > > >> [email protected]. > > >> > > To unsubscribe from this group, send email to > > >> [email protected]<google-appengine%[email protected]><google-appengine%2Bunsubscrib > [email protected]> > > >> . > > >> > > For more options, visit this group athttp:// > > >> groups.google.com/group/google-appengine?hl=en. > > > > >> -- > > >> You received this message because you are subscribed to the Google > Groups > > >> "Google App Engine" group. > > >> To post to this group, send email to > [email protected]. > > >> To unsubscribe from this group, send email to > > >> [email protected]<google-appengine%[email protected]><google-appengine%2Bunsubscrib > [email protected]> > > >> . > > >> For more options, visit this group at > > >>http://groups.google.com/group/google-appengine?hl=en. > > > > > -- > > > -- > > > Jeff > > > > > -- > > > You received this message because you are subscribed to the Google > Groups > > > "Google App Engine" group. > > > To post to this group, send email to [email protected] > . > > > To unsubscribe from this group, send email to > > > [email protected]<google-appengine%[email protected]><google-appengine%2Bunsubscrib > [email protected]> > > > . > > > For more options, visit this group at > > >http://groups.google.com/group/google-appengine?hl=en. > > -- > You received this message because you are subscribed to the Google Groups > "Google App Engine" group. > To post to this group, send email to [email protected]. > To unsubscribe from this group, send email to > [email protected]<google-appengine%[email protected]> > . > For more options, visit this group at > http://groups.google.com/group/google-appengine?hl=en. > > -- You received this message because you are subscribed to the Google Groups "Google App Engine" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en.
