... allocation strategy in ext3 in order to improve fsck performance. This ... problem implemented in this patch is to cluster indirect blocks together ... and + * make sure the following blocks are also prefetched. + */ + ext3_clear_buffer_prefetch(first_bh ...

linux-kernel - Abhishek Rai - Nov 16 2007 - 01:02

... allocation strategy in ext3 in order to > improve fsck performance. This ... problem implemented in this patch is to cluster > indirect blocks together ... the test. > > Notation: > - 'vanilla': regular ext3 without any changes > - 'mc': metaclustering ...

linux-kernel - Andrew Morton - Nov 16 2007 - 03:02

... this problem implemented in this patch is to cluster > > indirect blocks together ... without any changes > > - 'mc': metaclustering ext3 > > > > Benchmark 1: Sequential write to ... they are > > allocated alongside data blocks. The read path is unaffected ...

linux-kernel - Abhishek Rai - Nov 16 2007 - 18:27

I have implemented a revised patch that addresses the concerns raised ... (see ext3_fill_super). + * + * A note on ext3 metaclustering: + * + * Start of End of ... first try to allocate indirect + * blocks in the metacluster. */ + if (test_opt ...

linux-kernel - Abhishek Rai - Dec 21 2007 - 10:15

... I have implemented a revised patch that addresses the concerns raised ... first try to allocate indirect + * blocks in the metacluster. */ + if (test_opt( ... -2.6.23mm1-clean/fs/ext3/bitmap.c 2007-10-17 ...

linux-kernel - Abhishek Rai - Jan 10 2008 - 17:17

... and have fixed it in the new patch. Regarding some of ... do more aggressive prefetching of indirect blocks to help random reads ... so if metaclustering gets accepted in Ext3, I'll probably release ... cluster, thus > having a whole group of double index blocks in ...

linux-kernel - Abhishek Rai - Jan 13 2008 - 01:06

... a smart e2fsck that is doing clustering could just arrange to > pre-read ... of doubting metcluster size for unrelated ext3/fsck performance anomalies. The user should ... level and so on. Indirect block allocation proceeds in the opposite direction starting ...

linux-kernel - Abhishek Rai - Nov 16 2007 - 20:25

... the extents fit the inode, the metadata cluster area wouldn't be necessary at ... of doubting metcluster size for unrelated > ext3/fsck performance anomalies. I'm not ... a level and > so on. Indirect block allocation proceeds in the opposite direction > ...

linux-kernel - Theodore Tso - Nov 16 2007 - 22:58

... a level and > > so on. Indirect block allocation proceeds in the opposite direction > > ... is almost the same as regular ext3 performance but 1% worse than the ... the block group is an obvious choice. > > > In traditional metaclustering, once we run out ...

linux-kernel - Abhishek Rai - Nov 17 2007 - 04:58

... favor of for-real Ext3 hackers (you know who you are:-) ... we await review > but the patch has all its tabs replaced ... cluster, thus having a whole group of double index blocks in ... I tried hard. Barring mistakes in my estimates the sequential improvement ...

linux-kernel - Daniel Phillips - Jan 12 2008 - 02:05

... have room for more data blocks > in that blockgroup? It does ... smart e2fsck that is doing clustering could just arrange to pre ... if that disk has been in continuous create/delete/truncate/append ... the ratio of indirect blocks to data blocks will be different ...

linux-kernel - Theodore Tso - Nov 16 2007 - 17:11

... group. This keeps things simple and results in low fragmentation. However, if data blocks and their indirect blocks were to go ... > only for large files that span multiple block groups. > > Just an idea: > > What about putting it ...

linux-kernel - Abhishek Rai - Jan 11 2008 - 10:49

... have a large number > > of indirect blocks. For disks which have ... a huge speedup there. But in all cases, metaclustering doesn't ... if that disk has been in continuous create/delete/truncate/append ... the line "unsubscribe linux-kernel" in the body of a message ...

linux-kernel - Matt Mackall - Nov 16 2007 - 03:37

... 1.08s sys 1.98s) patch total runs 4 avg 10.50 ... does better than vanilla except in a few cases. Thanks, Abhishek ... have a large number > > > of indirect blocks. For disks which have ... if that disk has been in continuous create/delete/truncate/append ...

linux-kernel - Abhishek Rai - Nov 18 2007 - 11:52

... the blockgroup which > are used for indirect blocks. > > Presmably it starts around 50% ... asked about placement of the reserved blocks. Placement at the beginning of the ... at worst marginally less performance and in some cases better performance. I ...

linux-kernel - Andreas Dilger - Nov 16 2007 - 07:28

... Putting metacluster at the end of the block group gives slightly > inferior sequential read throughput compared to putting it in the > beginning or the middle, but ... only for large files that span multiple block groups. Just an idea: What about ...

linux-kernel - Bodo Eggert - Jan 11 2008 - 10:12

... . Your results are very impressive. In my opinion, the sooner this ... changes a significant amount of Ext3 code outside of the metacluster ... subject line like "Make Ext3 fsck way faster". Diff the patch ... ask Andrew to put it in -mm when you feel ready ...

linux-kernel - Daniel Phillips - Jan 11 2008 - 13:05

... user 1.08s sys 1.98s) > > patch total runs 4 avg 10.50 ... metaclustering does better than vanilla except in a few cases. > I think ... or is it dependent on other ext3 patches? Since I tested it ... the line "unsubscribe linux-kernel" in the body of a message ...

linux-kernel - Kyungmin Park - Nov 19 2007 - 06:34

... Your results are very impressive. In my opinion, the sooner this ... changes a significant > amount of Ext3 code outside of the metacluster ... subject line > like "Make Ext3 fsck way faster". Diff the patch ... ask Andrew to put it in -mm when you feel > ready ...

linux-kernel - Andrew Morton - Jan 11 2008 - 20:04

... Like 1gb, 1cpu and 400gb of a single disk? - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please ...

linux-kernel - John Stoffel - Nov 20 2007 - 16:25