geometric memory growth / mail delays

[Pascal Brisset]

Ulf Wiger (AL/EAB) writes:
 > Interestingly, they arrived in reverse order.

According to the SMTP headers, the delays occured in
"mailgw4.ericsson.se (Symantec Mail Security)".
Maybe this thing has a LIFO policy ?

Since most messages are not delayed, we should assume that it
explicitly manages a backlog, processing new emails as fast as
possible, but going through the queued ones as a background task
with lower priority.  That would be the right design choice if
the QoS is defined in terms of a threshold on the processing time.
This could be an idea for a new variant of gen_server :-)

Anyway we've had delays in France too (unrelated to erlang-questions),
so this is probably just a consequence of the latest virus outbreak.

-- Pascal