Yes, I know that first (and a few in case of firing them one after another in short delays like 200 ms like in my benchmark) request is slow, but I care about the next requests, that take ~50 ms without any distance between server and client and ~200 ms from appspot.com, 150 ms more then by pinging it. Well, sometimes I get a few requests in middle with ~600 ms lag - can it be because requesting my app every 200 ms still isn't enough to keep it warm?? I rather suspect that too big frequency of requests is making a bottleneck on client side.
I deleted all Aral Balkan's stuff and made my own code: http://2.latest.lagstest.appspot.com and this time tiny sourcecode: http://2.latest.lagstest.appspot.com/lagtest.zip and used URLLoaded instead of AMF - no difference. I tested it on a cheap PHP server I use and it was better, but when I test it know to make sure, it's much worse then GAE. Probably You can't achieve better (smaller) lags with this technology. But that is an answer I want to get from some guru, so I don't need to bother doing various benchmarks. Because lags increase when user makes client CPU busy, I'm pretty sure the 50 ms local lag is a flash issue. I will repost it on some flash forum. Still, I'm VERY suprised that noone before benchmarked GAE's lag... On 15 Lip, 01:38, Nick Joyce <[email protected]> wrote: > Sorry for the double post, couldn't see my first answer :) > > Appspot likes to consume instances of your application fairly quickly > if you don't get a lot of traffic. Initialising a cold one (for the > first request) can take quite a bit longer than a warm instance, due > to all the code loading you have to do (python imports etc.). You may > be hitting into this. > > I would suggest removing your dependancy on Aral's code and build an > app as described inhttp://pyamf.org/tutorials/gateways/appengine.html. > This will be able to tell you if that code is slowing you down or not. > > Hth, > > Nick > > On Jul 14, 4:49 am, Broadsmile <[email protected]> wrote: > > > > > Thanks for Your answer! Aral Balkan's example is very old and I'm > > aware of that. You suprised me by the technological jump of pyAMF, but > > unfortunately upgrading it has no effect in my case (probably because > > I don't send/get a lot of data, so pyAMF can't show off it's > > serialization/deserialization skills. > > > OLD:http://0.latest.lagstest.appspot.com/gra > > NEW:http://1.latest.lagstest.appspot.com/gra > > > Probably 50 ms offline lag is a fault of flash. But still, without > > these 50 ms lag server will have 150 ms, while it pings with 50. Maybe > > there's some multi wrong techniques stacked one on eachother. I will > > never know that until I see some better lag tester, that accomplishes > > ~50 ms (well, any under 150 ms will be nice) online lag. > > > So I repeat my questions and add more of them: > > 1. Does anyone have similar GAE lag tester? I don't care about > > technology, I care if it shows different pings. > > 2. Ad 7: does anyone know if there's something more in Aral Balkan's > > example that's ancient and wrong nowadays? > > 3. Does the God exist? > > 4. Is there any flash group I could ask it as well? (I'm kinda not a > > group discussant type and I'm not oriented...) > > > Thanks for Your try, Nick! And I'm looking forward to next hints! > > > On 13 Lip, 08:52, Nick Joyce <[email protected]> wrote: > > > > On Jul 11, 11:01 pm, Broadsmile <[email protected]> wrote: > > > > > Hi! I'm new to the GAE Engine, I found it will be a nice engine for my > > > > mmorpg game. For now I'm still investigating what Can I do with this > > > > technology, and what I can't. I want my game to be real-time based and > > > > am now testing the server to see how responsive it can be. > > > > > One of first things I did was pinging the app-name.appspot.com to see > > > > the lags. It was, and still is ~50ms. > > > > However! Pinging may be faster then actually sending and receiving > > > > data. So I made a simple benchmark in flash + > > > > python:http://lagstest.appspot.com/gra > > > > > I get about 200 ms lag, and it's too bad for some really fast action > > > > in realtime (althought I even don't really want so fast action so I > > > > don't need small lag). Sometimes lag is spiking to even a second and > > > > that's alarming. Also something weird happened - I get ~50 ms lag on > > > > offline devserver - that means, that flash, server or pyAMF is a > > > > significant factor! :O > > > > > I made the source code publichttp://lagstest.appspot.com/lagtest.zip > > > > (99% of the source code is by Aral Balkan - MIT licence - anyway). > > > > > I live in Poland. > > > > > So I have several questions: > > > > 1. Should I expect smaller lag? How to achieve that with flash + GAE? > > > > 2. Should I repost on Python GAE? > > > > 3. Any tips for a begginer ? > > > > 4. Any opensource projects to see, that have a good connection? > > > > 5. Is my place of living a factor? What do You see on this benchmark? > > > > 6. What's the sense of life? > > > > 7. Is my code allright or Aral Balkans code is already outdated? (I > > > > didn't upgrade AMF to the up-to-date version yet) > > > > > Thank You in Advance. > > > > I took a brief look at the zip and noticed that you are using PyAMF > > > 0.3.0b, which is ancient! :) > > > > There has been a lot of effort to optimise PyAMF since that version > > > (x4-x8 speed up in d/encoding is not unusual). If speed is your king > > > then I would suggest you take a look at the soon to be released 0.6 > > > [1] which is largely backwards compatible with all older versions. > > > > Cheers, > > > > Nick > > > > [1] -http://github.com/hydralabs/pyamf -- 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.
