[exim] ratelimit and smtp_accept_max_per_connection

Hello,

I'd like to ask your advice. What would be the effect of implementing
ratelimit instead of smtp_accept_max_per_connection?

I have smtp_accept_max_per_connection set and it happens that when it is
exceeded and the sending party is an MTA relaying emails through exim,
the connection is terminated, the MTA (Mercury for Windows) panics and
bounces some of the emails which were stopped because connection was
terminated. It may be that this is Mercury's fault (it should not bounce
but re-queue them) but I have no control over Mercury's behaviour and
wonder if I may be a bit safer by specifying ratelimit. However, if the
only result of ratelimit will be to terminate the connection silimary as
in the case of smtp_accept_max_per_connection, then there is not much
point in implementing it in my case.

I'd appreciate your advice. I just need to understand what are the
operational differences between ratelimit and
smtp_accept_max_per_connection.

Thanks!

Zbigniew Szalbot