[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Persistence
From: |
Jonathan S. Shapiro |
Subject: |
Re: Persistence |
Date: |
Mon, 31 Oct 2005 09:24:03 -0500 |
On Mon, 2005-10-31 at 08:04 +0100, R. Koot wrote:
> William Grim wrote:
> EROS makes a snapshot every 10 minutes, so if the system crashed 9
> minutes after the last snapshot you would, after a reboot, have the
> system in the state it was 9 minutes before the crash. The only way I
> can see to avoid these 'time jumps' would be making a snapshot on every
> IPC calls, but I don't think that will be feasable performance wise.
Actually, we checkpoint every 5 minutes, but this is tunable, and the
performance is reasonable at 100 sec. intervals.
For comparison, UNIX is no better. How often do you think that bdflush
runs?
shap