Date: Mon, 05 Apr 2004 10:24:20 +0200 From: Andy Sporner <sporner@nentec.de> To: Michael McDonald <m.mcdonald@computer.org> Cc: freebsd-cluster@freebsd.org Subject: Re: Request for Cluster Recommendations Message-ID: <407117B4.2000800@nentec.de> References: <002401c419e7$76692ac0$2f01a8c0@MICHAELIWZHLNY> <20040404113055.GA2677@ipx20050.ipxserver.de> <002c01c41aa5$1fbd6b50$2f01a8c0@MICHAELIWZHLNY>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi Michael, >For the most part, my codes have been data >parallel and have invlolved broadcasting >parameters and merging results at the end of >distributed serial computations. They have >involved many evaluations of small matrices >and data sets and I get better speedups with >simultaneous serial executions. Much of the >work I'm looking towards would make use >of a grid approach in the style of seti@home >or the factoring projects. > > Seti@home is very nice indeed. I am a participant as well (Captain Blank) >I would expect bursts of communication >separated by periods of computation; >Overall, communication wouldn't be >so much of a bottleneck, but I'd like it >to be fast when it does occur. If Andy >Sporner's 30% figure holds up, I think >local disks as buffers would allow the >network access to be smeared out. I've got >40% in mind as an upper performance limit >for ethernet due to collisions, but I can't back >that up. Local buffers seem to allow for >scheduling comm. so as to avoid collisions. > With a switch you wouldn't normally have enough collisions to worry about, unless you approach the bandwidth of the media. When I spoke of the 30% I was refering to a extended burst. We benchmark our firewall loadbalancer(nitro) with 4 firewalls and a traffic generation farm for entire weekends for reliability testing, this is where I observed this value. The Firewall machines are 2.8 Ghz Athlon servers. > >Starting out with 100Mbit may make sense as >a cheaper/simpler startup - no NIC >purchases. After running some simulation >and benchmarking with the apps, upgrading to Gigabit wouldn't be an undue burden. > That's not a bad idea, but many of the 1U servers (not to mention the better motherboards (such as ASUS) are with GB as standard). The switch is a little more expensive, but not that drastic. >Some initial attempts at estimating >communication & compute demands would >be in order. Aside from the fiber options, is the cabling the same for 100Mbit and Gbit? > Using Cat6 -- yes, but we have also used cat 5 with some success. If you start wth CAT-6 you will always win. Good luck! Andy
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?407117B4.2000800>