Thursday, 10 July 2008

Sarah just got her University results for semester 1.

A distinction and a credit.

That was starting the semester recovering from heart surgery. I'm very proud of her.

She's doing three subjects next semester, and then she's finished first year. Pretty impressive progress really. I know I wouldn't stand a snowflake's chance in hell of doing well studying externally.

I'm in the process of converting my poor-man's SAN from IDE disks attached to the host by IDE-USB adapters to SATA disks attached by eSATA to SATA cables.

I had some spare time tonight, so I thought I'd prepare the new RAID-10 on brutus. The intent is to then take the 4-port USB 2.0 card out of minotaur during a lull in recordings on Saturday, and put it in brutus as well, and do a giant pvmove, then put the 4-port eSATA card back in minotaur with the new disks attached.

So I've got the 4-port eSATA card (a Silicon Image SiI 3124) in brutus, and I've built the RAID-10 with mdadm and it's busily reconstructing.

I'm getting a reconstruction speed of around 60Mb per second (/proc/mdstat says 60254K/sec right now). I have no idea if that's good, bad, or indifferent.

I do know that output of hdparm -tT is very nice by comparison:

apollock@brutus:~$ sudo hdparm -tT /dev/sda

/dev/sda:
 Timing cached reads:   260 MB in  2.01 seconds = 129.06 MB/sec
 Timing buffered disk reads:  222 MB in  3.01 seconds =  73.77 MB/sec

compared to

apollock@minotaur:~$ sudo hdparm -tT /dev/sda

/dev/sda:
 Timing cached reads:   162 MB in  2.02 seconds =  80.20 MB/sec
 Timing buffered disk reads:   48 MB in  3.04 seconds =  15.78 MB/sec

Those times are with a reconstruction running on brutus and a recording being written to the array on minotaur, so not the most controlled benchmarking environment.

If I can sit on my hands long enough, I'll try some benchmarks from the MythTV machine, comparing the performance of the SAN before and after switching disks.