switches
Many of your questions may have already been answered in earlier discussions or in the FAQ. The search results page will indicate current discussions as well as past list serves, articles, and papers.
Martin Siegert siegert at sfu.caThu Jan 4 15:08:13 PST 2001
- Previous message: Scyld, RedHat 7.0 and Athlon's
- Next message: 2nd Cluster Computing in the Sciences Conference
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Hello everybody, I'm planning to expand my cluster to ~ 100 processors with 3 x 100BaseT channel bonding. Obviously, I need switches - unfortunately I don't know very much about these beasts. Here is my idea: I'll buy 48 dual processor nodes with 3 NICs each (at least 4 in the master node). Hence, I'll need 3 48 port switches. Can anybody recommend a particular 48 port (or more) switch (best performance, lowest price, as usual :-). My web search didn't come up with much: Cisco Catalyst 2948G Cisco Catalyst 3548XL Enterasys Vertical Horizon VH-4802 Intel Express 420T Did I miss anything? Which can be recommended? Do they all have backplanes that support 4.8Gb/s? I guess in this price range all of these should allow forcing to 100BaseT/full-duplex (I don't trust autonegotiation - should I?). There are other modular switches (e.g., Dlink DES-5600, HP Procurve 4000M), but these seem to be more expensive, right? Alternatively, I could use stackable 24 port switches, however then I need a 2.4Gb/s link between the stacked switches (full duplex). I don't know whether I can get that. 48 port switches seem to be easier in any case. Thanks for your help and comments in advance. Cheers, Martin ======================================================================== Martin Siegert Academic Computing Services phone: (604) 291-4691 Simon Fraser University fax: (604) 291-4242 Burnaby, British Columbia email: siegert at sfu.ca Canada V5A 1S6 ========================================================================
- Previous message: Scyld, RedHat 7.0 and Athlon's
- Next message: 2nd Cluster Computing in the Sciences Conference
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Beowulf mailing list
