> Given that just one ethernet link can easily transport 100MB/s and a > typical SATA hard drive can only do 70MB/s how is the ethernet a > bottleneck? And I use LACP to bond together 2. So I have 2.5 hard > drives worth of bandwidth. But that's just raw throughput. We all know > that the seeks are what usually slow you down. Each of my AoE servers > have 2G of RAM in them for caching.
Do you have any iozone, or some other benchmark to support that? I have always had the understanding that even a single local disk can nearly sustain 1Gbit/sec, with bursts that are higher thanks to cache, while over the network, due to network overhead, the most you can get is about 800Mbit from a 1Gbit connection... Especially when that 1Gbit connection is shared by a bunch of machines... Here is one typical 500Mb 7200rpm sata disk, which is able to sustain over 1Gbit/sec throughput: http://www.hitachigst.com/portal/binary/com.epicentric.contentmanagement.ser vlet.ContentDeliveryServlet/hgst/products/Deskstar/p7k500/HGST_Deskstar_P7K5 00_DS_FINAL.pdf Local disk would be even faster if you're using some kind of RAID mirroring or striping. PS. 2G of ram in a server doesn't sound like much to me. I never buy any laptop with less than 4G anymore. > Temporary disk IO? Will they all be doing this temporary disk IO at the > same time? I have at least 200MB/s of bandwidth minimum between any > compute node and any storage node. That's a lot of IO for most > applications. And given that this can all be spread out over a number > of disks on the SAN side of things plus the cache in the SAN head the > IOPS can really get up there. Yes, all the compute nodes will be working in parallel, that's the whole point. They will all be writing to local scratch space, and consuming a lot of cpu and ram at the same time. _______________________________________________ Tech mailing list [email protected] http://lopsa.org/cgi-bin/mailman/listinfo/tech This list provided by the League of Professional System Administrators http://lopsa.org/
