Gentoo Archives: gentoo-user

From: Uwe Thiem <uwix@××××.na>
To: gentoo-user@l.g.o
Subject: Re: [gentoo-user] LTSP vs. Diskless Nodes
Date: Sat, 29 Oct 2005 21:05:45
Message-Id: 200510292255.24822.uwix@iway.na
In Reply to: Re: [gentoo-user] LTSP vs. Diskless Nodes by Bob Sanders
1 On 29 October 2005 20:14, Bob Sanders wrote:
2 > Since I'm rambling now, guess I should do the rest of the memory
3 > download...
4
5 Let me join you in rumbling. ;-)
6
7 Before I start just a little background: I do quite some consulting for
8 SchoolNet Namibia (http://www.schoolnet.na) which has hooked up about 250
9 schools all over the country. All of them (well, almost all - some externally
10 sponsored ones run Windows) running linux thin clients connecting to a linux
11 server running the South African linux distribution OpenLab
12 (http://www.getopenlab.com) and LTSP.
13
14 Typical server configuration:
15 P4 2.8 GHz
16 2GB ram
17 2 x 80GB normal IDE harddrive
18
19 Typical client:
20 refurbished x86 box
21 currently PII or PIII
22 32MB ram
23
24 >
25 > One of the big problems with Linux diskless is it really doesn't scale
26 > well, it doesn't allow for clients to run multiple versions of the os,
27
28 Why would you want to do that?
29
30 > nor
31 > for different arch types to co-exist off one server of a different arch
32 > type.
33
34 That's possible, of course. The boot prom on the NIC in the client can boot
35 any kernel/X server under /tftpboot on the server.
36
37 >
38 > Additionally, a typical diskless setup exports /usr as a read only file
39 > system (which by most definitions, it is supposed to be).
40
41 A typical LTSP server doesn't export /usr at all. There is no need for it. The
42 client runs a kernel and an X server. If you want local devices to work, it
43 also needs to run some other small daemons. All *applications* run on the
44 server.
45
46 [ snip ]
47
48 > Another not so well thought out diskless problem with Linux is all the
49 > setups use one kernel under /tftpboot or, at least the Gentoo Diskless
50 > guide uses /diskless, which makes it a bit movable, but then falls into
51 > showing the path as - /diskless/xxx.xxx.xxx.xxx (an IP number) instead of
52 > using the node name. The other problem, is they all assume only one kernel
53 > vs. a kernel for each host.
54
55 Not necessarily. The boot prom on the client's NIC decides what to load from
56 under /tftpboot on the server.
57
58 My experiences with LTSP so far show: With a server like mentioned at the
59 begin and fast ethernet, up to 20 clients are working well if you don't allow
60 too graphics-intensive apps like movie players or that type of games. For
61 more clients (up to 40), you need more ram on the server and a Gb connection
62 between the server and the switch (clients can remain on 100Mb ethernet, of
63 course).
64
65 For small businesses, I prefer a different solution that involved solid state
66 clients that boot from non-volatile ram. In that case, the client is
67 completely independent of the server. All they talk to each other is X.
68
69 Cheers from the beginning southern African summer
70 Uwe
71 (It's starting to rain outside, and I am feeling *good*!)
72
73 --
74 95% of all programmers rate themselves among the top 5% of all software
75 developers. - Linus Torvalds
76
77 http://www.uwix.iway.na (last updated: 20.06.2004)
78 --
79 gentoo-user@g.o mailing list

Replies

Subject Author
Re: [gentoo-user] LTSP vs. Diskless Nodes Jerry McBride <mcbrides9@×××××××.net>
Re: [gentoo-user] LTSP vs. Diskless Nodes Ryan Viljoen <ravilj@×××××.com>
Re: [gentoo-user] LTSP vs. Diskless Nodes Bob Sanders <rmsand@××××××××××.net>