On Sun, Aug 9, 2009 at 3:39 PM, Alex Clemesha <cleme...@gmail.com> wrote:
> > > If sagenb.org is to slow it is *not* a hardware issue -- we have enough > > hardware to support a huge number of users. It's also likely not a > > bandwidth issue because Univ of Washington has tons of bandwidth. It's a > > software issue: > > (1) the notebook server software itself needs to be made more > scalable > > (2) the notebook server runs using the free VMware server, which is > very > > limited (e.g., at most 8GB RAM and at most 2 virtual processors). > > > > I will be address both (1) and (2) in October. Regarding (2), I'll > probably > > switch from vmware to VirtualBox, and (1) will just take a massive amount > of > > hard work and careful benchmarking. > I hope that you'll be looking very closely at the architecture > of codenode, as to not duplicate effort, because we already have > put in a massive amount of hard work making it scalable and efficient. Well my first step will be to write automated user simulation code to make precise in practice how scalable and efficient sagenb.org is. I'll also run the code on codenode to see if indeed codenode is any more scalable and efficient. For all I know codenode is way better than the Sage notebook in theory and way worse in practice. I'll find out the truth and post some data/benchmarks/tables. That's step 1. > > Probably the most important areas where the Sage notebook and > codenode differ, and thus aid codenode in being more scalable and > efficient are: > > - The components are very decoupled. You can run the backend as > a totally separate process as the frontend (a Twisted process). > You also can run the frontend in very optimized separate processes > (e.g. run Django > from Nginx's WSGI, and serve all static files from Nginx, which is > extremely fast). As you know, the Sage notebook front end and backend are also totally separate processes. I don't understand why you think this is a difference. How they communicate is different -- Sage uses pexpect and codenode doesn't. > > - The entire architecture of codenode is based around a relational > database, > as opposed to doing file+shell operations, and using things like pickle. > Once you have many users, with many notebooks, plus revision > history, plus search, etc this becomes a huge deal. With Django's > ORM you can use it simply and locally with a database like SQLite > and when it comes time to scale out you can use a big MySQL > instance running on a dedicated machine. That's definitely a difference. I would actually prefer if the Sage notebook used flat files more than it already does, though it currently uses them much more than it did when you were involved. I don't care whether one technology sounds way better than another -- what I care about is benchmarks and actual usability and maintainability. I'm not making any claims about the Sage notebook currently being better than anything though. Just that I don't intend to approach improving matters via "theoretical best practices", but instead via benchmarking. Well anyways, you know I'm as lazy as possible, and that I wish I didn't have to do anything at all with respect to making the Sage notebook much better. So whatever there is that I can benefit from in codenode, I'll be interested. -- William > > > -Alex > > > > p.s: > For those not familiar with codenode (http://codenode.org) > and are interested in code Notebooks in general I'd strongly > suggest getting on our mailing list > here:codenode-de...@googlegroups.com<here%3acodenode-de...@googlegroups.com> > We been making some great progress recently, but more input > and collaboration could never hurt :-) > p.p.s: > William and I basically "invented" the Sage Notebook at UCSD > around 2006. codenode was started about 1 1/2 later in hopes > of creating a better, more generalized code Notebook. > > > > > >> > >> We use a dualboot install based on the KNOPPIX DVD and I think they have > >> iceweasel. > >> > > > > Sage should work fine with iceweasel. > > > >> > >> > >> Thanx, > >> A. Jorge Garcia > >> email > >> mailto:calcp...@aol.com > >> website > >> http://calcpage.tripod.com > >> weblog > >> http://calcpage.tripod.com/shadowfax > >> > >> Teacher & Professor > >> Applied Mathematics, Physics & Computer Science > >> Baldwin Senior High School & Nassau Community College > >> > >> In a message dated 8/9/2009 1:21:10 A.M. Eastern Daylight Time, > >> docfleetw...@gmail.com writes: > >> > >> It is great to hear (read) that the online site is designed to remain > >> indefinitely. I think that is the way I will go with my students. > >> > >> ________________________________ > >> > > > > > > > > -- > > William Stein > > Associate Professor of Mathematics > > University of Washington > > http://wstein.org > > > > > > > > > > > -- > Alex Clemesha > clemesha.org > > > > -- William Stein Associate Professor of Mathematics University of Washington http://wstein.org --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-support@googlegroups.com To unsubscribe from this group, send email to sage-support-unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-support URLs: http://www.sagemath.org -~----------~----~----~----~------~----~------~--~---
