ITP: postal - SMTP benchmark
I am currently writing a SMTP benchmark program I'm calling "postal". I aim
to have a test release out by the end of the week and hopefully have a 1.0
release (which will include POP testing) out by the end of the year.
The main feature of note is that the way the program will manage multiple
addresses.
If you have a smallish test machine (say 64M of RAM) or a machine that has a
slow hard drive (no RAID) then managing a list of 30,000,000 users for
testing a mail server will be impractical/impossible.
What postal will do is to have strings of the following form:
XXXX ...[a-z][0-9]
Which means that any user-name from the user-file which matches the regular
expression XXXX will have it's 4th character replaced with a random letter
from [a-z] and it's 5th character replaced with a random letter from [0-9].
If this example was how your test mail server was setup then the amount of
storage for the list of test users would be 1/260th of the size it would
otherwise be.
This has worked really well on previous testing software I've written, but
unfortunately the client didn't want me to GPL it. So I'm writing it again
(and better). Also I'll release the code to expand user-names seperately
(hopefully within a few days) so that anyone else who's writing test software
can use it.
One thing that I want to do is use a normal distribution for the sizes of the
test emails. If anyone has a suggestion on how to calculate this (or some
GPL code that I can use) then I would appreciate it greatly.
Also I would like to know what the average sizes of emails are. I plan to
write some scripts to extract data from mail log files (if there aren't
already such scripts) and then summarise it. If anyone has any such scripts
already then I would appreciate being mailed a copy of them. If anyone is
running moderate sized email servers (10,000 or more users) and would like to
run such scripts and send me the output then I would appreciate that too.
NB I don't think that a script which says how many emails were received in
the time period and the number of messages in each binary multiple of sizes
is an invasion of privacy for anyone.
--
Electronic information tampers with your soul.
Reply to: