There are other posts about regular boot times, so how about the worst case, file system checks. That's that check that runs every 30 boots or so just because we are paranoid about making sure the file system is sane.
| Boot | Intrepid (est) | Jaunty with ext4 |
| Normal | 30 – 45 seconds | 18 – 22 seconds |
| File System Check | 5 – 10 minutes | 43 seconds |
Of course a lot of this is due to the use of the new ext4 file system (ext3 will be the default in Jaunty again because we are paranoid). At 43 seconds I don't mind the file system check nearly as much. Progress. :)
6 comments:
How does this chart even make sense? Either give us Jaunty ext3 times, or intrepid ext4 times... the way you've presented the data, it seems as if you're looking for praise for the ext4 people than the ubuntu devs.
I can't really give Intrepid ext4 boot times (at least not easily). Both the ext4 devs and the ubuntu devs helped make the boot faster, so I am praising both.
http://kernelnewbies.org/Ext4
quote:
"Fsck is a very slow operation, especially the first step: checking all the inodes in the file system. In Ext4, at the end of each group's inode table will be stored a list of unused inodes (with a checksum, for safety), so fsck will not check those inodes. The result is that total fsck time improves from 2 to 20 times, depending on the number of used inodes (http://kerneltrap.org/Linux/Improving_fsck_Speeds_in_Ext4). It must be noticed that it's fsck, and not Ext4, who will build the list of unused inodes. This means that you must run fsck to get the list of unused inodes built, and only the next fsck run will be faster (you need to pass a fsck in order to convert a Ext3 filesystem to Ext4 anyway). "
In other words, run fsck a few times and it should be faster.
Those numbers make perfect sense. Interpid didn't offer ext4 and Jaunty does. That would be like saying it's not fare to compare Ubuntu with the "Big M" because Microsoft doesn't use ext4.
Those numbers make perfect sense. Interpid didn't offer ext4 and Jaunty does. That would be like saying it's not fare to compare Ubuntu with the "Big M" because Microsoft doesn't use ext4.
Problem with ext4 is that no sane person would use it for a production system, now that Theodore tried hard to destroy his and the filesystem's reputation during the write-back affair.
So praising performance improvements which require ext4 doesn't make much sense that the points. It's the same like drooling about performance improvements of some imaginary rndfs:
ssize_t
rndfs_write (int fd, const void *buf, > size_t *count)
{
/* This freaking clever algorithm > gives
* an amazing performance boost of 500%!!! **drool**
*/
if (rand() % 5)
return count;
return rndfs_real_write (fd, buf, count);
}
Let's hope ext4's reputation can be restored.
taschenorakel,
It looks like tytso finally got the point in the current 2.6.29 thread on lkml. I'm not sure whether he can make ext4 work both fast and reliable but it seems several key kernel guys will be working on the problem:
http://marc.info/?t=123785113100004&r=1&w=2
Post a Comment