[Opensim-dev] Concerning recent grid instability andinconsistency

[kino at daxtron.com]

On bandwidth issues everyone might want to check p. 5-8 and 13 of a Cisco presentation
"Virtual Environments and their impact on the Network"
http://xianrenaud.typepad.com/weblog/files/TechTalk2.pdf

If you think of each client as a channel, then its 70 kBps per visiting client or channel.  
That's why hosting on a small box at a co-lo is can work for some and can put an upper bound on visitors to a home ADSL box even if you have an ultra-high end CPU, since there is the pattern is backwards from ADSL assumptions. Instead of home box input >> home box output for web surfing, Opensim has home box output >> home box input, and the output is what clients depend upon.

Bests,
Kino Coursey

http://www.daxtron.com

-----Original message-----
From: "Teravus Ovares" teravus at gmail.com
Date: Thu,  6 Mar 2008 14:29:48 -0500
To: opensim-dev at lists.berlios.de
Subject: Re: [Opensim-dev] Concerning recent grid instability
	andinconsistency

>  From what I've read, it can assuming you don't have regions load balancing
>  between machines that are across the internet.
>  However, this brings up another interesting topic..    generally Is a cheap
>  DSL connection going to be enough to host a 'ton' of users anyway?    You'd
>  think you'd have to want to host a ton of users to get any benefit from
>  using the load balancing.
>  
>  Best Regards
>  
>  Teravus
>  
>  
>  On 3/6/08, Grumly <grumly57 at hotmail.com> wrote:
>  >
>  >  Would this explain why DSL home hosted adjacent sims crossing is so shaky
>  > sometimes ?
>  > In this case, I am afraid such simulators could also not take advantage of
>  > the new load balancing feature.
>  >
>  >  *From:* Michael Wright <michaelwri22 at yahoo.co.uk>
>  > *Sent:* Thursday, March 06, 2008 12:47 PM
>  > *To:* opensim-dev at lists.berlios.de
>  > *Subject:* Re: [Opensim-dev] Concerning recent grid instability
>  > andinconsistency
>  >
>  >
>  > I haven't had much time to follow this discussion, but just a quick note.
>  > We used to have a REST like communications protocol between the regions, but
>  > this lead to a lot of problems. Granted it was like the rest of the backend
>  > protocols we use, very err primitive. But I just not sure http/xml is the
>  > best protocol to be used there, as a lot of the calls are time critical. And
>  > there are some things we are yet to add which would increase the network
>  > traffic between the regions by quite a bit. Thats why we switched to
>  > remoting so we could at least have a binary protocol. Maybe we do need to
>  > switch to udp, like I believe Linden Labs do (or at least they used to).
>  >
>  > *Brian McBee <heartwide at gmail.com>* wrote:
>  >
>  > Coming late to this discussion, but i have a couple of comments about our
>  > use of remoting in the interregion communications:
>  >
>  > 1. It seems like remoting is really designed for a scenario where you own
>  > both the client and server ends of the connection, and have a reasonable
>  > expectation that the server is up. In a distributed grid environment like
>  > OSGRID is running, that assumption isn't true.
>  >
>  > 2. When playing with the teleport code, I tried to find a way to lower the
>  > timeout on the remoting calls, since a failed teleport took forever to
>  > resolve. It appears that, at least the way we are doing it, the timeout is
>  > hardcoded and cannot be changed. It looks like Microsoft tries to provide a
>  > way to change that timeout, but it doesn't actually work.
>  >
>  > Perhaps longer term, we should be looking at replacing remoting with a
>  > REST interface like we are using for the sim to grid communications.
>  >
>  >
>  > _______________________________________________
>  > Opensim-dev mailing list
>  > Opensim-dev at lists.berlios.de
>  > https://lists.berlios.de/mailman/listinfo/opensim-dev
>  >
>  >
>  >  ------------------------------
>  > Rise to the challenge for Sport Relief with Yahoo! for Good<http://us.rd.yahoo.com/mailuk/taglines/isp/control/*http://us.rd.yahoo.com/evt=51947/*http://uk.promotions.yahoo.com/forgood/>
>  >
>  > ------------------------------
>  >
>  > _______________________________________________
>  > Opensim-dev mailing list
>  > Opensim-dev at lists.berlios.de
>  > https://lists.berlios.de/mailman/listinfo/opensim-dev
>  >
>  >
>  > _______________________________________________
>  > Opensim-dev mailing list
>  > Opensim-dev at lists.berlios.de
>  > https://lists.berlios.de/mailman/listinfo/opensim-dev
>  >
>  >
>