Postfix bottlenecks my ISP for big mailouts

[webxchris]

When we send an e-mail to a large number of addresses (5k max), our internet dies. We go from having steady 30ms ping to outside IPs to 900+ and timeouts. Basically we lose internet due to overload of outgoing e-mail.

I notice whenever we send out a mailing, the Postfix Mail Queue gets about 10-20 groups (large groups) of e-mail addresses with copies of the e-mail to go out. It seems to get stuck at that point. I can mark all as Hold and about a minute later our internet goes back to normal.

This is not a server resource issue, just an ISP bandwidth issue. I'm running on a standard cable connection so I am somewhat limited. (Speed test shows average of 25 Mbps download, 5.4 Mbps upload)

I have already attempted to tweak the Postfix settings to resolve this problem, after reading the documentation. It doesn't seem to have had any affect. I wanted to see if I am missing something or if someone knows the solution to my problem off the top of their head.

Here are my settings that may be related:

initial_destination_concurrency = 2
default_destination_concurrency_limit = 4
local_destination_concurrency_limit = 2
smtp_destination_concurrency_limit = 4
smtpd_sasl_auth_enable = yes
smtpd_sasl_security_options = noanonymous
broken_sasl_auth_clients = yes
smtpd_recipient_restrictions = permit_mynetworks permit_sasl_authenticated reject_unauth_destination
mailbox_size_limit = 0
allow_percent_hack = no
home_mailbox = Maildir/
message_size_limit = 100240000
default_destination_recipient_limit = 2
default_destination_concurrency_limit = 4
default_process_limit = 50
fork_delay = 5s
deliver_lock_delay = 5s
qmgr_message_active_limit = 10000
maximal_backoff_time = 4000s
smtp_rcpt_timeout = 100s
smtp_data_init_timeout = 40s
smtp_data_done_timeout = 40s
smtp_quit_timeout = 40s
smtp_connect_timeout = 20s
fallback_relay =
smtp_helo_timeout = 30s
smtpd_recipient_limit = 250
smtp_destination_recipient_limit = 1000
queue_run_delay = 600s
minimal_backoff_time = 1000s
smtp_mail_timeout = 100s

I appreciate any help!

[Rubberman]

Either establish a QOS (quality of service) rule in your router that will throttle the speed at which mail is sent, or some other throttling mechanism. You are suffering from network congestion because the mail is taking all the bandwidth you have - hence the increased ping times and general timeout failures. There is no magic elixir for this except possibly a dedicated pipe to the internet for your outgoing mail. In any case, I'd definitely put it on a separate nic on the mail server so that regular traffic isn't trying to move over the same physical link.