1 |
Am 06.09.2011 21:24, schrieb James Broadhead: |
2 |
> On 6 September 2011 19:55, Permjacov Evgeniy <permeakra@×××××.com> wrote: |
3 |
>> On 09/06/2011 09:26 PM, Pandu Poluan wrote: |
4 |
>>> Disk I/O Characteristic: Occasional writes during 'normal' usage, |
5 |
>>> once-a-week eix-sync + emerge -avuD |
6 |
>>> Priority: Stable (i.e., less chance of corruption), least CPU usage. |
7 |
> |
8 |
> You would have to profile this, but I imagine that the best approach |
9 |
> would be to compile in a RAM disk and copy. I think that you're |
10 |
> probably trying to optimise the wrong part of this problem. |
11 |
> |
12 |
> As for ext3/ext4, the improvements to fsck alone make ext4 the FS of |
13 |
> choice between the two. |
14 |
> |
15 |
> JB |
16 |
> |
17 |
|
18 |
Pandu is building a firewall. Putting a ton of RAM in it just for the |
19 |
sake for system updates is plain overkill and -- depending on his IaaS |
20 |
provider -- pretty expensive. |
21 |
|
22 |
Regards, |
23 |
Florian Philipp |