Raid 5 performance

From: Todd Lewis (todd_at_qdsdirect.com)
Date: 02/08/04

  • Next message: Uwe Doering: "Re: Raid 5 performance" 
    Date: Sat, 07 Feb 2004 21:57:50 -0600
To: freebsd-performance@freebsd.org

    I am using FreeBSD 4.9, with a 3ware RAID 5
    1 gig memory 2.8g p4

    Three questions.

    1. FreeBSD has a 16k block size. The RAID card is set at 64k
        Block size(its sweet spot). My logic tells me that
        increasing the block size to 64k would increase disk
        read and write access. But, everything I read suggest
        going above 64k is dangerous. Are their any recomendations
        on performance a stability concerns when increasnig the
        block size to 64k when using a RAID controler.

    2. The vfs.hirunningspace variable defualts to 1meg. From what I've
        read this looks like a buffer. I'm guessing that its set to
        1meg becuase most drives have 1~2 megs of memory. So following
        that logic and with safety in mind. For drives with 4 megs
        cache, I would set vfs.hirunningspace to 2 megs. 8megs of cache
        4 megs to vfs.hirunningspace. So, my 64 megs raid control would
        have a vfs.hirunningspace 32.
    3. Any other stellar performance info would be appriciated.

    netstat -m

    mbuf usage:
            GEN list: 4/480 (in use/in pool)
            CPU #0 list: 443/656 (in use/in pool)
            Total: 447/1136 (in use/in pool)
            Maximum number allowed on each CPU list: 512
            Maximum possible: 51200
            Allocated mbuf types:
              358 mbufs allocated to data
              89 mbufs allocated to packet headers
            2% of mbuf map consumed
    mbuf cluster usage:
            GEN list: 12/564 (in use/in pool)
            CPU #0 list: 263/304 (in use/in pool)
            Total: 275/868 (in use/in pool)
            Maximum number allowed on each CPU list: 128
            Maximum possible: 25600
            3% of cluster map consumed
    2020 KBytes of wired memory reserved (32% in use)
    0 requests for memory denied
    0 requests for memory delayed
    0 calls to protocol drain routines

    dd if=/dev/da0s1f of=/dev/null bs=1m count=100

    code:
    100+0 records in
    100+0 records out
    104857600 bytes transferred in 6.331798 secs (16560478 bytes/sec)

    _______________________________________________
    freebsd-performance@freebsd.org mailing list
    http://lists.freebsd.org/mailman/listinfo/freebsd-performance
    To unsubscribe, send any mail to "freebsd-performance-unsubscribe@freebsd.org"

  • Next message: Uwe Doering: "Re: Raid 5 performance"

    Relevant Pages

    • Re: Database performance
      ... Memory: Depending on your processing needs, if a small search set can ... you can and should by all means run ext2 on your scratch ... Different RAID levels are appropriate to different tasks. ... Multiple controllers help. ...
      (Debian-User)
    • Re: New Lightwave Workstation
      ... Id imagine it's better to install a PCI Sata raid ... RAID, in which parity data to reconstruct drive data ... drives can fail simultaneously and can be replaced without data loss. ... total addressible memory and that has to account for all memory addresses in ...
      (comp.graphics.apps.lightwave)
    • Re: Can anyone recommend a MB?
      ... I really don't want to clear 2 RIMMS to gain half a gig of memory. ... ATI Radeon X800 Pro - AGP ... AGP, 4xDDR, PCI slots, USB2 ports. ... Look for ICH5R, for Southbridge RAID. ...
      (alt.comp.periphs.mainboard.asus)
    • SBS 2003 Migration to new Hardware
      ... I've just bought a new Compaq ML370 with Raid one Scsi ... Lots of memory, dual processers, full redundency all ... Plus I doubt the old machine would take the migration ... user info, groups and permissions and exchange info, ...
      (microsoft.public.windows.server.sbs)
    • RE: Where is the performance bottleneck?
      ... memory and main memory can't even go that fast. ... to compare sw raid to a hw raid you should count every byte of ... decreases the cache effectiveness and you end up paying later after the sw ... send the line "unsubscribe linux-kernel" in ...
      (Linux-Kernel)