Small Project Starter Suggestions
I've got an old server and some machines not in use that could be useful for data-logging and internet access in a science lab. The machines don't have CD-ROMs so I'm thinking I either end up sticking a cd-rom in each machine to build it, or get the workstations to act as dumb terminals connect to the server - I have a 100mb switch and the machines have Intel pro cards in so should be acceptable speed. Any pointers on a good way to go forward, what to read and where to start? -- Glenn Cameron King Henry VIII School, Coventry
On Tuesday 23 Oct 2001 19:20, Glenn Cameron wrote:
I've got an old server and some machines not in use that could be useful for data-logging and internet access in a science lab. The machines don't have CD-ROMs so I'm thinking I either end up sticking a cd-rom in each machine to build it, or get the workstations to act as dumb terminals connect to the server - I have a 100mb switch and the machines have Intel pro cards in so should be acceptable speed. Any pointers on a good way to go forward, what to read and where to start?
I don't 100% understand what you want to do with them. If it's just simple 'net access, and perhaps word processing/whatever, I'd do an install over the network, and just use them as standalone machines, although of course going over the LAN for 'net access. -- Cheers, Chris Howells -- chris@chrishowells.co.uk, howells@kde.org Web: http://chrishowells.co.uk, PGP key: http://chrishowells.co.uk/pgp.txt KDE: http://www.koffice.org, http://edu.kde.org, http://usability.kde.org
On Tuesday 23 October 2001 19:20, Glenn Cameron wrote:
I've got an old server and some machines not in use that could be useful for data-logging and internet access in a science lab. The machines don't have CD-ROMs so I'm thinking I either end up sticking a cd-rom in each machine to build it, or get the workstations to act as dumb terminals connect to the server - I have a 100mb switch and the machines have Intel pro cards in so should be acceptable speed. Any pointers on a good way to go forward, what to read and where to start?
We are currently looking into the possibility of getting the finance to make a Linux thin client data-logging utility. This would enable your older machines to not only use data logging but run up to date Office aplications such as Star Office 6 off a server and also Internet access etc. Unfortunately, given the nature of funding applications it will probably take a year or so to get this to completion. Depending on the machines, you could build a thin client network but you would need a reasonably beefy server, say £800 on the hardware. Also it is really necessary to have at least 1 meg of video RAM in the client machines and 2 meg is better, in order to run the X graphical interface. 10 meg ethernet to a 100 meg switch and then to the server should be more than enough. -- IanL
On Tue, 23 Oct 2001, Ian Lynch wrote:
On Tuesday 23 October 2001 19:20, Glenn Cameron wrote:
I've got an old server and some machines not in use that could be useful for data-logging and internet access in a science lab. The machines don't have CD-ROMs so I'm thinking I either end up sticking a cd-rom in each machine to build it, or get the workstations to act as dumb terminals connect to the server - I have a 100mb switch and the machines have Intel pro cards in so should be acceptable speed. Any pointers on a good way to go forward, what to read and where to start?
We are currently looking into the possibility of getting the finance to make a Linux thin client data-logging utility. This would enable your older machines to not only use data logging but run up to date Office aplications such as Star Office 6 off a server and also Internet access etc. Unfortunately, given the nature of funding applications it will probably take a year or so to get this to completion.
In that case don't forget to size your software to work on next year's obsolete machines, not todays.
Depending on the machines, you could build a thin client network but you would need a reasonably beefy server, say £800 on the hardware.
Also it is really necessary to have at least 1 meg of video RAM in the client machines and 2 meg is better, in order to run the X graphical interface. 10 meg ethernet to a 100 meg switch and then to the server should be more than enough.
If you're really building a data logger then I can't see the need to run an X server. Maybe X *clients*? But probably something simpler, even a little HTTP server, would do the job. If you really want a desktop then you're building desktop machines that can do data logging, which IMHO is a different thing; the unpredictable activity produced by the desktop clients would tend to make the timing problem harder. For me the trickiest parts of doing a data logger would be interfacing to the sensors and perhaps getting a stable enough timebase. I've driven parallel I/O such as printer ports from userspace quite successfully (polling rates around the 1000Hz mark) from Tcl, and I'm sure Perl etc. would do the job too, and provide an HTTP interface, on a very small box. Bob G
On Wed, 24 Oct 2001, Robert J Gautier wrote:
Depending on the machines, you could build a thin client network but you would need a reasonably beefy server, say £800 on the hardware. Also it is really necessary to have at least 1 meg of video RAM in the client machines and 2 meg is better, in order to run the X graphical interface. 10 meg ethernet to a 100 meg switch and then to the server should be more than enough. If you're really building a data logger then I can't see the need to run an X server. Maybe X *clients*? But probably something simpler, even a little HTTP server, would do the job. If you really want a desktop then you're building desktop machines that can do data logging, which IMHO is a different thing; the unpredictable activity produced by the desktop clients would tend to make the timing problem harder. For me the trickiest parts of doing a data logger would be interfacing to the sensors and perhaps getting a stable enough timebase. I've driven parallel I/O such as printer ports from userspace quite successfully (polling rates around the 1000Hz mark) from Tcl, and I'm sure Perl etc. would do the job too, and provide an HTTP interface, on a very small box.
We will be doing the development work for the data logging and control part of this project. The plan is to create tiny servers to run on the machines physically connected to the data logging equipment, communicating with a selection of front-end programs via TCP/IP. We are aiming to create a standard interface to a variety of data logging equipment so that the front-ends can use any type of temperature sensor without having to be aware of all the different makes, models and protocols that the various bits of educational data-logging kit use. For the control side, we plan to extend Squeak since it already provides a near-perfect environment for manipulating objects graphically, building programs using drag-and-drop, etc. Only problem is processor load: it requires something of the order of a 700MHz CPU to run a single front-end. Fortunately, this applies only to the computer running the front-end - in the case of a thin-client system this would be the central server and the client workstations would be more than capable of running the tiny servers that actually talked to the hardware. Michael
participants (5)
-
Chris Howells
-
Glenn Cameron
-
Ian Lynch
-
Michael Brown
-
Robert J Gautier