Messages in this thread | | | Date | Wed, 12 May 1999 14:32:28 -0400 (EDT) | From | Steve Willer <> | Subject | Re: PATCH: rewritten bdflush |
| |
On Tue, 11 May 1999, Zack Weinberg wrote:
> There should be no change in what bdflush actually does. I've been > running 2.2.7-as-patched for 13 hours and it works fine; there was a > six-hour compile job in that interval. I'm sure others can stress it > harder though, I'd be interested in results.
I made a separate posting of this, but this one will have actual data in it.
Your patch doesn't seem to scale up to heavy writes, like you might find in a transaction-oriented database. Running without update, the bdflush doesn't seem to be able to keep up with the I/O load (without making me very nervous about recoverability and the age of buffers). Ideally, the writes are done at a pace that's equal to a per-30-second sync, but spread over the full 30 seconds. In other words, the ideal is what update used to do. :-)
Here's the output of 'vmstat 2' during a collection of uniformly heavy inserts, with update running as 'update -S -s 30' (so I can be sure of the test parameters). Hopefully this will illustrate my point. Note how user time drops: this is a result of the bursty I/O.
procs memory swap io system cpu r b w swpd free buff cache si so bi bo in cs us sy id 1 0 0 3628 8208 186092 10484 0 0 0 0 1134 42 95 5 0 2 0 0 3628 8192 186056 10484 0 0 1 0 1419 33 92 6 2 2 0 0 3628 8160 186088 10484 0 0 0 250 1648 58 86 8 6 2 0 0 3628 8324 186032 10484 0 0 1 0 896 12 91 3 6 2 0 0 3628 8148 186168 10484 0 0 1 250 1058 13 92 7 0 2 0 0 3628 8216 186120 10484 0 0 0 0 1191 19 95 5 0 2 0 0 3628 8156 186120 10484 0 0 1 0 1169 23 95 4 1 2 0 0 3728 8172 186204 10484 0 50 1 263 1112 51 93 5 2 2 0 1 3748 7928 186436 10484 0 10 0 3 1274 29 94 6 0 2 0 1 4112 8084 186648 10484 0 182 1 296 1397 121 90 8 2 2 0 0 4364 7768 187300 10484 0 126 0 32 1226 86 95 5 0 2 0 1 4396 5756 189152 10484 0 16 1 4 586 17 78 5 18 2 0 0 4452 8144 186876 10484 0 28 0 507 1055 26 92 8 0 2 0 0 4452 8188 186836 10484 0 0 1 250 1083 17 88 5 7 2 1 0 4452 8188 186780 10484 0 0 1 1520 1385 98 89 9 2 1 1 0 4452 8220 186748 10484 0 0 0 4599 1429 877 91 7 1 1 2 0 4452 8224 186744 10484 0 0 1 5390 1644 1043 76 11 13 2 0 0 4452 8112 186812 10484 0 0 0 128 706 29 88 5 6 2 0 0 4452 8172 186760 10484 0 0 1 0 896 12 96 3 0 2 0 0 4452 8284 186752 10484 0 0 0 500 1016 15 94 6 0 2 0 0 4452 8144 186880 10484 0 0 1 0 1087 18 95 4 1 2 0 0 4452 8248 186776 10484 0 0 1 0 1168 27 92 6 1 2 0 0 4452 8244 186776 10484 0 0 0 250 1333 18 96 4 0 1 0 0 4452 8236 186780 10484 0 0 0 0 1435 31 93 5 2 2 0 0 4452 8180 186836 10484 0 0 0 250 2013 115 84 8 7 2 0 0 4452 8276 186740 10484 0 0 0 0 1847 94 84 8 9 1 0 0 4452 8244 186772 10484 0 0 0 0 2068 102 86 9 6 2 0 0 4452 8184 186832 10484 0 0 1 250 1543 43 85 8 7
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |