The 09-02 billing report is in, and it looks correct, with 293.12 frontend instance hours (down from 09-01's 764.94 or previous days' ~3000). If you were to lower to max-idle-instances=1, that would be reduced another 96 instance hours (down to <200), but you'll want to keep an eye out that latencies are still tolerable if you do that.
I hope that clears things up. Let me know if it doesn't or you have other questions here. On Mon, Sep 5, 2011 at 2:50 PM, Millisecond <millisec...@gmail.com> wrote: > Jon, > > Thank you for the detailed response and taking the time to look at my > specific application. > > I'll take a look at the new billing statement in the morning (I'm > central european time) and see what I can tell. > > > On Sep 5, 10:49 pm, Jon McAlister <jon...@google.com> wrote: >> Hi millisecond, I can illuminate some of the issues here. >> >> As Johan said, the bill will be based on active-instances [the orange >> line] plus max-idle-instances. That is correct. >> >> I can see though how your 09-01 billing report is confusing though, I >> can explain that. Also, I'm very sorry that the billing reports are >> delayed so long, we're working on that. >> >> The billing reports are computed on the usage for the day, based on >> Pacific Standard Time. In your case, the setting of >> max-idle-instances=5 was made at 2011/09/01-04:28:41. So, it applied >> to the remaining 19.5 hours of the day, but not to the initial 4.5 >> hours for the day. This is why it didn't drop as low as expected (it >> dropped from ~3000 instance-hours to 764 instance hours, but obviously >> it should be going lower than that). But, we can figure out the >> average billed instance rate for the latter 19.5 hours of the day by >> solving for: >> >> 3000 / 24 * 4.5 + N * 19.5 = 764 >> >> Which is N=10.3. As such, I would expect the 09-02 billing report to >> have ~250 instance hours. We'll see in a few hours if this is correct >> or not. >> >> The number 10.3 seems correct though. Your active-instances rate graph >> is a sawtooth graph with a trough of 3 and a peak of 13. So, the >> average of that graph, plus 5, coming out to 10.3, looks correct to >> me. >> >> I'll check back in on you once the 09-02 billing report comes out in a >> few hours for you. >> >> >> >> >> >> >> >> On Mon, Sep 5, 2011 at 7:12 AM, Johan Euphrosine <pro...@google.com> wrote: >> > The Max Idle instances slider help text states: >> >> > *You will not be charged for instances over the specified maximum* >> >> > So if you are, it is a billing bug and you should ask a refund. >> >> > Feel free to a production issue with your application id if you want >> > us to track the instance inconsistencies for your application: >> >http://code.google.com/p/googleappengine/issues/entry?template=Produc... >> >> > I would also suggest to fill a feature request for faster billing >> > preview/reporting: >> >http://code.google.com/p/googleappengine/issues/entry?template=Featur... >> >> > Hope that helps. >> >> > On Mon, Sep 5, 2011 at 4:00 PM, Millisecond <millisec...@gmail.com> wrote: >> >> I agree that we shouldn't be, it just seems that we are. >> >> >> I'll just have to wait for more days to show up in Billing History >> >> where I have screenshots of the dashboard available to compare >> >> against. >> >> >> Frustrating to have to wait 5 days to see what something is going to >> >> be billed at. We have basically 2 cycles to make changes and see the >> >> effects before billing goes into effect and the two things biting us >> >> are instance-hours and datastore writes. One is totally opaque and >> >> the other appears to be inconsistently reported in the live graphs >> >> (right now I have "20 Total" in text, 35 total / 12 active in the >> >> graph). >> >> >> On Sep 5, 3:40 pm, Johan Euphrosine <pro...@google.com> wrote: >> >>> You should never be charged for more than: Active Instance + Max Idle >> >>> instances, even if the scheduler keeps more than Max Idle instances >> >>> around. >> >> >>> Setting Min-Pending-Latency to 60ms instruct the scheduler to wait at >> >>> least 60ms if all the instances are busy before deciding to spawn a >> >>> new instance for handling an incoming request. >> >> >>> You can maximize existing instance usage over new instance creation by >> >>> increasing that value, but this could come at the expense of >> >>> increasing request latency. >> >> >>> Hope that helps. >> >> >>> On Mon, Sep 5, 2011 at 3:12 PM, Millisecond <millisec...@gmail.com> >> >>> wrote: >> >>> > It looks like we are being charged at the higher rate of 30-35 >> >>> > instance hours / hour even though only ~10 are active at a time and >> >>> > max-idle is set to 5. Although it's still hard to tell as we're >> >>> > behind ~5 days in billing history summaries. >> >> >>> > Is the current scheduler going to be changed before new-billing is >> >>> > implemented? Seems like a must. >> >> >>> > On Sep 1, 4:48 pm, Millisecond <millisec...@gmail.com> wrote: >> >>> >> The table at the top of the dashboard is showing 23 active instances, >> >>> >> QPS of 2.6 and a latency of 481ms. Which makes some sense as I'm >> >>> >> hovering between 40 and 80 QPS overall. >> >> >>> >> I've set the scheduler to have max 5 idle instances and the min >> >>> >> latency to 60ms, but when I pull down the graph to the "Instances" >> >>> >> display, it has me hovering between ~5 and ~12 active and 30-35 >> >>> >> total. >> >> >>> >> Is the graph just not accurate and we're billed for what's in the text >> >>> >> area? Are we charged for active instances or total instances or >> >>> >> instances in the text area? Why isn't it more-or-less "active + 5 as >> >>> >> max = total"? >> >> >>> >> With a 3-4 day delay on my billing history reporting, this is going to >> >>> >> be very hard to tweak if I can't see correct numbers on the dashboard. >> >> >>> > -- >> >>> > 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 >> >>> > athttp://groups.google.com/group/google-appengine?hl=en. >> >> >>> -- >> >>> Johan Euphrosine (proppy) >> >>> Developer Programs Engineer >> >>> Google Developer Relations >> >> >> -- >> >> 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 >> >> athttp://groups.google.com/group/google-appengine?hl=en. >> >> > -- >> > Johan Euphrosine (proppy) >> > Developer Programs Engineer >> > Google Developer Relations >> >> > -- >> > 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 >> > 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 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. > > -- 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.
