[FreeNX-kNX] Minimum bandwidth req's

[Igor Manassypov]

Well,

the bw hogger in my case is pdf browsing by the remote clients. As I
suspected, pdf pages are being transferred as plain images and that
drags me down. I am too on rogers line (384kbps up), and even one client
becomes extremely slow. 
Would you be able to suggest a better way for text presentation through
nx tunnel rather than pdf? Would html offer any speed advantage?

> Message: 1
> Date: Thu, 22 Feb 2007 18:39:42 +0100
> From: "Jo-Erlend Schinstad" <joerlend.schinstad at gmail.com>
> Subject: Re: [FreeNX-kNX] Minimum bandwidth req's
> To: "User Support for FreeNX Server and kNX Client"
> 	<freenx-knx at kde.org>
> Message-ID:
> 	<bc66a2e50702220939j64efaa63j8283365012848a2f at mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> 
> On 22/02/07, Igor Manassypov <imanassypov at cicore.com> wrote:
> > >From your experience, what are the minimum bandwidth requirements for
> > near-realtime response from nxserver? I cant figure that part out, I
> > need to calculate how many concurrent clients can a particular line
> > sustain.
> >
> >
> > Thanks!
> >
> >
> > -igor
> 
> Wow, that was one large post. Well, a friend of mine told me he was
> using nx via gprs, which is about 30kbps, I think. It's abit difficult
> to answer your question, other than the standard "it depends". Because
> it does, depend on what you'll be using nx for. Nx uses advanced
> caching and compression techniques, which makes nx really fast. Do you
> use audio? One way, or both ways? That requires bandwidth. Do you only
> use applications with more or less static appearance? That won't
> require alot of bandwidth, because it's cached on the client. I've
> used tree concurrent connections on a 384kbps upstream line, and that
> went very well. That was without audio. I could probably have added at
> least two clients without problems. If I were you, I'd monitor the
> clients bandwidth usage while connected to the server. Do the things
> your users will be doing and see for yourself. Divide your available
> bandwidth by the peak bandwidth usage from the test, and you'll have
> the smallest number of clients you can have.
> 
> Good luck :)
> Jo-Erlend