2009/11/29 Eric <shel...@gmail.com>: > Thanks for all your comments. > Regarding Python/Java speed, 99% of the runtime is spent iterating in > an attempt to converge to some numerical solution . > Loops, arithmetic and memory updates. I would guess an interpreted > language like Python (am I right?) would be much slower. >
You can try a simulation test before your decision. > About the instance set up time, I agree that if query frequency > suddenly rises from 100/sec to 200/sec the system I suggested would > stall, but I'm don't think I'm expecting such drastic changes in a > small amount of time. > Regarding my suggestion on new instance creation, don't I have a say > on when and how that happens? Or is it totally up to AppEngine to > define when a new instance is created? Can I, for example, have 10 > instances that are simply waiting for situations like Keakon > described? At least not right now, It's up to the app master. http://code.google.com/events/io/2009/sessions/FromSparkPlugToDriveTrain.html > I'm trying to form some intuitive picture of how GAE works by guessing > and hoping to be corrected by someone when/if I'm wrong. > Maybe I require a cloud platform that allows tinkering at lower > levels? > > On Nov 28, 9:54 pm, Andy Freeman <ana...@earthlink.net> wrote: >> > Maybe I don't understand something, but why should the 5 second setup >> > on a new instance bother me? A new instance should be created when >> > other instances are near capacity, and not when they exceed it, right? >> > So once initialized it can be "dummy-run" internally and only >> > available 5 seconds later while the existing instance continue to take >> > care of the incoming queries. >> >> What makes you think that the request that causes the creation of a >> new instance doesn't wait for the creation of said instance? (The >> scheme you suggest is plausible, but why do you think that it's how >> appengine works?) >> >> On Nov 28, 9:03 am, Eric <shel...@gmail.com> wrote: >> >> > Thanks for the response. >> >> > Maybe I don't understand something, but why should the 5 second setup >> > on a new instance bother me? A new instance should be created when >> > other instances are near capacity, and not when they exceed it, right? >> > So once initialized it can be "dummy-run" internally and only >> > available 5 seconds later while the existing instance continue to take >> > care of the incoming queries. >> >> > Also, do you think the latency requirements are realistic with GAE? >> > That is, in the ordinary case, could the response be consistently >> > served back to the querying user with delay of max 3 seconds? >> >> > On Nov 28, 8:35 am, 风笑雪 <kea...@gmail.com> wrote: >> >> > > The White House hosted an online town hall meeting on GAE with GWT, >> > > and received 700 hits per second at its >> > > peak.http://google-code-updates.blogspot.com/2009/04/google-developer-prod... >> >> > > But more than 1000 queries a second is never been tested. >> >> > > I think Java is not a good choice in your case. When your user >> > > suddenly increasing, starting a new Jave instance may cost more than 5 >> > > seconds, while Python needs less than 1 second. >> >> > > 2009/11/27 Eric <shel...@gmail.com>: >> >> > > > Hi, >> >> > > > I wish to set up a CPU-intensive time-important query service for >> > > > users on the internet. >> > > > Is GAE with Java the right choice? (as compared to other clouds, or >> > > > non-cloud architecture) >> > > > Specifically, in terms of: >> > > > 1) pricing >> > > > 2) latency resulting from slow CPU, JIT compiles, etc.. >> > > > 3) latency resulting from communication of processes inside the cloud >> > > > (e.g. a queuing process and a calculation process) >> > > > 4) latency of communication between cloud and end user >> >> > > > A usage scenario I am expecting is: >> > > > - a typical user sends a query (XML of size around 1K) once every 30 >> > > > seconds on average, >> > > > - Each query requires a numerical computation of average time 0.2 sec >> > > > and max time 1 sec (on a 1 GHz Pentium). The computation requires no >> > > > data other than the query itself. >> > > > - The delay a user experiences between sending a query and receiving a >> > > > response should be on average no more than 2 seconds and in general no >> > > > more than 5 seconds. >> > > > - A background save to a DB of the response should occur (not time >> > > > critical) >> > > > - There can be up to 30000 simultaneous users - i.e., on average 1000 >> > > > queries a second, each requiring an average 0.2 sec calculation, so >> > > > that would necessitate around 200 CPUs. >> >> > > > Is this feasible on GAE Java? >> > > > If so, where can I learn about the correct design methodology for such >> > > > a project on GAE? >> >> > > > If this is the wrong forum to ask this, I'd appreciate redirection. >> >> > > > Thanks, >> >> > > > Eric >> >> > > > -- >> >> > > > 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-appeng...@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.-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-appeng...@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-appeng...@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.
