13 Apr
2011
13 Apr
'11
4:10 a.m.
On Tue, Apr 12, 2011 at 11:45 PM, Henk-Jan van Tuyl
If you want to know what the last thing was that your application was doing, before it crashed (e.g. at the customers site), you better write every message immediately to disk.
Yes, I suppose it would depend on your use case whether you wanted to
optimize for throughput or for latency to secondary storage. I just
know from experience that for high-throughput server applications,
writing each message to disk is going to introduce unacceptable
context-switching overhead.
G
--
Gregory Collins