| 1 |
-----BEGIN PGP SIGNED MESSAGE----- |
| 2 |
Hash: SHA256 |
| 3 |
|
| 4 |
On 17/01/14 11:27 AM, Michał Górny wrote: |
| 5 |
> Hello, all. |
| 6 |
> |
| 7 |
> I'm using squashfs to hold my Gentoo repositories on all of my |
| 8 |
> systems for some time. As you probably know, this allows me to save |
| 9 |
> space while keeping portage fast. However, it makes updating the |
| 10 |
> tree quite burdensome and time-consuming. |
| 11 |
> |
| 12 |
> We're already hosting daily gx86 tarballs on our mirrors, and |
| 13 |
> deltas made using diffball. Those can be used with Zac's |
| 14 |
> emerge-delta-webrsync to get daily updates done with minimal |
| 15 |
> network overhead. Sadly, it takes the whole process even more time |
| 16 |
> consuming :). |
| 17 |
> |
| 18 |
> Therefore, I'd like to suggest an alternative solution that could |
| 19 |
> help out Gentoo users that use squashfs for gx86 and would like to |
| 20 |
> be able to get daily updates fast and easy. |
| 21 |
> |
| 22 |
> The idea is to host -- along with the tarballs -- daily squashfs |
| 23 |
> images of gx86 in a chosen format. Additionally, the images would |
| 24 |
> come with deltas made using xdelta3 or a similar tool. Those deltas |
| 25 |
> -- with a slight download overhead -- would allow very fast |
| 26 |
> updates of the squashfs. |
| 27 |
> |
| 28 |
> Now some numbers. I did some tests 'converting' late gx86 daily |
| 29 |
> tarballs to squashfs. I've used squashfs 4.2 with LZO compression |
| 30 |
> since it's quite good and very fast. |
| 31 |
> |
| 32 |
> 96M portage-20140108.sqfs 96M portage-20140109.sqfs 96M |
| 33 |
> portage-20140110.sqfs 96M portage-20140111.sqfs 96M |
| 34 |
> portage-20140112.sqfs 96M portage-20140113.sqfs 97M |
| 35 |
> portage-20140114.sqfs 97M portage-20140115.sqfs |
| 36 |
> |
| 37 |
> For deltas, I've used xdelta3 with max compression (-9) and djw |
| 38 |
> secondary compression (it gave ~0.1M smaller files than fgk and |
| 39 |
> ~0.5M gain than with no secondary compression). |
| 40 |
> |
| 41 |
> 4,9M portage-20140108.sqfs-portage-20140109.sqfs.vcdiff.djw 6,3M |
| 42 |
> portage-20140109.sqfs-portage-20140110.sqfs.vcdiff.djw 5,6M |
| 43 |
> portage-20140110.sqfs-portage-20140111.sqfs.vcdiff.djw 8,9M |
| 44 |
> portage-20140111.sqfs-portage-20140112.sqfs.vcdiff.djw 6,3M |
| 45 |
> portage-20140112.sqfs-portage-20140113.sqfs.vcdiff.djw 7,8M |
| 46 |
> portage-20140113.sqfs-portage-20140114.sqfs.vcdiff.djw 8,5M |
| 47 |
> portage-20140114.sqfs-portage-20140115.sqfs.vcdiff.djw |
| 48 |
> |
| 49 |
> As you can see, the deltas are quite large compared to the actual |
| 50 |
> changes. However, we could have expected that since we're diffing a |
| 51 |
> compressed filesystem. What's important, however, is that applying |
| 52 |
> it takes ~2.5 second on my 2 GHz Athlon64. |
| 53 |
> |
| 54 |
> So, even with the extra download time, the update is much faster |
| 55 |
> than recreating the squashfs. And unlike some types of unionfs, it |
| 56 |
> doesn't come with extra runtime slowdown. |
| 57 |
> |
| 58 |
> What do you think? |
| 59 |
> |
| 60 |
|
| 61 |
PLEASE DO! This sounds fantastic, and is something i've been |
| 62 |
considering proposing for some time. |
| 63 |
-----BEGIN PGP SIGNATURE----- |
| 64 |
Version: GnuPG v2.0.22 (GNU/Linux) |
| 65 |
|
| 66 |
iF4EAREIAAYFAlLZWoMACgkQ2ugaI38ACPAb0gEAwFPWdtI5J8l9QH1YTrKe2Mbm |
| 67 |
OwPL4wGg9ORaHv0FkVcA/iLsP/z3uz3sJoWciR5ZCZ73HblyDgH/flAhakNhl3NZ |
| 68 |
=ool3 |
| 69 |
-----END PGP SIGNATURE----- |
Archives