At first I'd like to express to Andrey and Bill the high appreciation of the teachers in the NetMath network in the German land of Rhineland-Palatinate for providing and supporting SageCell as a free service. Usually it works fine and the support is fast and helpful. We understand that resources are limited. So the Virtual Campus of Rhinland-Palatinate has considered setting up its own SageCell server for the community, but found that it requires more resources than the Campus can currently afford. I'd suggest to maintain - at a central place - a list of *Known Issues/Tips and Tricks* listing problems that might occur and possible workarounds.
Am Donnerstag, 6. April 2017 07:09:39 UTC+2 schrieb Andrey Novoseltsev: > > On Wednesday, 5 April 2017 15:03:28 UTC-6, Paul Masson wrote: >> >> I would appreciate some clarity on this statement. If you have future >> restrictions in mind, I would like to know what that might entail. And >> before you impose restrictions of any sort, you need to have explicit >> guidelines publicly available detailing what is and what is not going to be >> allowed. >> > > To add some numbers to William's and David's responses (which I completely > support): > - Computing 1+1 via SageMathCell is drastically different from doing it in > JavaScript not only because it is done on a remote server, but because it > involves forking a dedicated Sage process. You don't have to wait for it, > typically, because there are some waiting for you, but this forking does > take noticeable time, something like 1/10 of a second. A page with 10 very > easy cells thus consumes 1 second. > - Each interact requires a dedicated Sage process running to support it, > which takes noticeable RAM, something like 100MB, so a page with 10 > interacts consumes 1GB. > - Americas are served by a machine with 4 CPU cores and 15GB RAM, so 15 > students opening a page with 10 interacts will consume all RAM. > > A sensible solution to this is to increase capacity on demand via preemt > instrances or containers or whatever, but a) it has to be implemented and > b) the current minimal setup already costs ~$130/month. So the actual > solution to overused server will be more likely to limit requests for the > same user/session to something reasonable. Say allow only one cell or > evaluation per second. Which should be just fine for someone who actually > reads a web-page. Those who want more resources and/or stability should > consider running their own servers as I have done in the past. > -- You received this message because you are subscribed to the Google Groups "sage-cell" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-cell/f17ce181-6f28-4f17-a335-3210b026093e%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
