The decision to tell the appliances to delete Spam was made yesterday. Anything coming in flagged as Spam, not Suspect Spam, will be dropped. This is 99% of the stuff flagged as spam, as 'suspect' is a really small category. This does reduce the load on the Exchange front-end servers as they have to do much less spam checking and handle a lot fewer messages. Though, as I'll show below, only a little less data.
And now, fun stats for Yesterday!
Total messages processed: 193,242
Percentage flagged as Spam: 49%
Percentage flagged as Suspect Spam: less than 1%
Virus mails: 731 messages
Top virus: Trojan.Peacomm (45% of viruses)
Top non-WWU inbound mailer: 129.41.62.246
Top spam sender: service@watermarkcu.org, 4% of spam (go phish!)
The mail flow goes something like this:
[inbound] -> BigIP -> Appliance -> BigIP -> Exchange FrontEnd -> Exchange
The BigIP is used to load-balance between the exchange front-ends for SMTP traffic. As it flows through the BigIP, I get stats on data volume over those ports
Mail volume to Appliances: 1.7G
Mail volume to Exchange: 1.4G
So data volume isn't greatly affected by dropping 49% of incoming mail. What is affected is the number of messages being processed. The front-end servers weren't terribly loaded as it was, this just means that Outlook Web Access is more responsive than it was.
And now, fun stats for Yesterday!
Total messages processed: 193,242
Percentage flagged as Spam: 49%
Percentage flagged as Suspect Spam: less than 1%
Virus mails: 731 messages
Top virus: Trojan.Peacomm (45% of viruses)
Top non-WWU inbound mailer: 129.41.62.246
Top spam sender: service@watermarkcu.org, 4% of spam (go phish!)
The mail flow goes something like this:
[inbound] -> BigIP -> Appliance -> BigIP -> Exchange FrontEnd -> Exchange
The BigIP is used to load-balance between the exchange front-ends for SMTP traffic. As it flows through the BigIP, I get stats on data volume over those ports
Mail volume to Appliances: 1.7G
Mail volume to Exchange: 1.4G
So data volume isn't greatly affected by dropping 49% of incoming mail. What is affected is the number of messages being processed. The front-end servers weren't terribly loaded as it was, this just means that Outlook Web Access is more responsive than it was.