RE: Antivirus checker mailet

[Vincenzo Gianferrari Pini <vi...@praxis.it>]

> Now, taking as example the James system I started in my company, there are
> about 220 users (not big but not small) and in the peak hours there are
> about 2 messages per minute going on; probably the load would be
> very low (I
> will write a small java routine executing virusscan in a loop
> scanning some
> attachments to get an idea of the potential cpu load).
>

I've written the small java routine and done the measurement, starting
several times in a loop McAfee's VirusScan on a directory with some
pre-loaded executables, on a 1.2GHz Pentium III CPU under Windows XP (my
laptop):
Two cases:
1) With 2 executables in the directory, for 537KB total, it took 258 ms per
execution.
2) With a third large executable in the directory, for 6.28MB total, it took
282 ms per execution.

Considerations:

1) The size and number of files to scan don't seem to be important.
2) As the files where pre-loaded and always the same, they were in cache and
so I didn't measure the I/O time related to write them to the directory.
3) I think at the end that the performance would not be so bad; we can take
also in account the fact that messages with attachments are not
statistically frequent, and in case we scan only attachments and not the
message body ...

What do you think?

Thanks,

Vincenzo


---------------------------------------------------------------------
To unsubscribe, e-mail: james-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: james-user-help@jakarta.apache.org

RE: Antivirus checker mailet

["Noel J. Bergman" <no...@devtech.com>]

Seems interesting, and a good start.  For now, you'd probably want to use a
file based spool, and point the scanner at the file, if it can handle it.

	--- Noel


---------------------------------------------------------------------
To unsubscribe, e-mail: james-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: james-user-help@jakarta.apache.org