From: Mike S. <oop...@gm...> - 2009-01-19 04:06:14
|
On Sun, Jan 18, 2009 at 7:55 PM, Rob Spearman <ro...@di...> wrote: > > Regarding DSS or other image hosting, I think it would be relatively > easy to get this donated. I don't think it's something we want to run > if we can help it, better to spend our time/resources elsewhere. Whom specifically should we ask? A server farm, a university, a research organization? Google, Facebook (they mirror some open-source projects), Microsoft, Novell? I'm just throwing out ideas. Mike > Is there a written license now? That was the hang up last time I looked > into this. The host would want some kind of mention. > > Thanks, > > Rob > > > On Sun, 2009-01-18 at 17:48 -0600, Mike Storm wrote: >> > I thought of using a peer to peer system at first, but it seems that >> > it's difficult to avoid huge latencies. In Stellarium, we need very >> > low latency for getting a lot of small images as fast as possible. >> >> I was thinking only a few constantly-connected, high-speed clients >> would be doing the sharing. For example, my friend down the hall has a >> custom-built gaming box that's always connected to the university >> network. He could even choose to run some sort of daemon, if he >> doesn't want Stellarium open all the time. These clients would be >> equivalent to supernodes in the Gnutella network, which I'm fairly >> familiar with. (Nobody actually uses a pure peer-to-peer network >> anymore, due to those latency problems you're worried about.) >> >> Perhaps a mixed system - when requesting an image tile, the leaf (end >> user) 1) queries its own cache, 2) queries the supernodes, 3) queries >> the master server. If in step 2 the supernodes don't ping back within, >> say, 50ms, go to step 3 and ask the master server. I think that this >> would eliminate our latency problems and substantially reduce load on >> the master server. Even if none of the supernodes respond, nobody's >> going to notice the 1/20th of a second it took to ask them. And >> end-users will never need to know that this is going on, unless they >> choose to be a supernode. Does this all make sense? >> >> I think I can prototype this. Thoughts? >> >> > >> > But I am not an expert for these questions though, any thoughs >> > welcome. In the mean time, we have enough money from donations to set >> > up at least 1 server for a couple of years. >> >> Wow! I didn't know that. By the way, what's the size of the entire DSS >> catalog we're using? And do we have any estimate of the bandwidth that >> it's going to take? >> >> Mike >> >> ------------------------------------------------------------------------------ >> This SF.net email is sponsored by: >> SourcForge Community >> SourceForge wants to tell your story. >> http://p.sf.net/sfu/sf-spreadtheword >> _______________________________________________ >> Stellarium-pubdevel mailing list >> Ste...@li... >> https://lists.sourceforge.net/lists/listinfo/stellarium-pubdevel > > > ------------------------------------------------------------------------------ > This SF.net email is sponsored by: > SourcForge Community > SourceForge wants to tell your story. > http://p.sf.net/sfu/sf-spreadtheword > _______________________________________________ > Stellarium-pubdevel mailing list > Ste...@li... > https://lists.sourceforge.net/lists/listinfo/stellarium-pubdevel > |