More importantly, what node is going to cache a 2 GB file? If you want to have soft size limits, ie having the node still passes a file of greater size then it will cache (I'm not sure about that, bandwidth is generally more valuable then diskspace), you could simply use a circular buffer of whatever size you can tolerate and then overwrite the beginning when you have reached the end (obviously making sure the stuff you overwrite is already sent).
On Fri, 04 Aug 2000, Timm Murray wrote: > > > > 4. How do you break up files in the datastore and the datastore > > > > indices to get around file size limits imposed by filesystems? > > > what? > > > > I think he's referring to the 2GB file size limit on Linux and similar > > limitations in other OSs. > > I apologize for the caps, but . . . > > WHOS GOING TO BE PASSING AROUND A 2GB FILE???!!! > > :) > > > > > > In fact, I think a lot of your criticisms really don't make much > > > sense. You're not designing a webserver. It doesnt *have* to scale to > > > enterprise levels. > > > > I was thinking about this. Whenever the datastore starts having > > scalability problems then that probably means we're not properly balancing > > the load throughout the network. Then again, it's possible that we may end > > up with so much traffic that even with proper load balancing there is > > still a need for scalable datastores. This would be the case if there were > > several large sites running a trusted subnetwork and swapping huge amounts > > of data between each other. Of course any large node could be replaced > > with several smaller nodes, but there may be times when running a large > > node is a sensible thing to do, such as when you just happen to have a > > big, fast computer with lots of HD space and bandwidth. You might as well > > put a big, fat node on it. In general, though, I neither see much of a > > need for highly scalable datastores, or any particular reason *not* to > > make the datastore highly scalable. It's more of an issue with a C port > > since that's what people trying to run high traffic nodes will probably > > run anyway. > > > > > > > > _______________________________________________ > > Freenet-dev mailing list > > Freenet-dev at lists.sourceforge.net > > http://lists.sourceforge.net/mailman/listinfo/freenet-dev > > > > > _______________________________________________ > Freenet-dev mailing list > Freenet-dev at lists.sourceforge.net > http://lists.sourceforge.net/mailman/listinfo/freenet-dev -- \oskar _______________________________________________ Freenet-dev mailing list Freenet-dev at lists.sourceforge.net http://lists.sourceforge.net/mailman/listinfo/freenet-dev
