Throughput

• Table of Contents
• What is the maximum sending speed achievable with your MTA software?
• Are there any limitations or factors that can affect sending speeds?
• Does this MTA have a hard cap on both inbound and outbound connections, and if so then what is it?

What is the maximum sending speed achievable with your MTA software?

We have benchmarked GreenArrow at over 25 million emails per hour on a single instance of GreenArrow (on older generation Intel Xeon hardware) - this document describes the parameters we were using in the test when that speed was achieved.

The MTA team at Klaviyo reported in an interview that they’d managed to send 50,000 emails per second (180 million per hour) across a cluster of 12 instances.

Are there any limitations or factors that can affect sending speeds?

Non-Enterprise GreenArrow Engine licenses have their sending speed limited to the contracted capacity. Enterprise licenses have no sending speed limit.

There are a number of factors that affect the sending speed:

• Server hardware resources and speed
• Message size
• Message acceptance rate
• Number of DKIM signatures and key length of DKIM signatures
• Use of Click & Open tracking in GreenArrow

Does this MTA have a hard cap on both inbound and outbound connections, and if so then what is it?

GreenArrow has configurable limits on the number of incoming and outgoing connections, and it is flexible enough to have no problems meeting the needs of high-volume senders.

For outgoing SMTP connections, these are the maximum limits:

• Maximum of 40,000 concurrent connections for first delivery attempts of messages.
• Maximum of 400,000 concurrent connections for subsequent delivery attempts of messages.

These limits are per-pod/per-instance, so a GreenArrow cluster with multiple pods/instances can greatly exceed these limits.

For incoming SMTP connections, there is no maximum limit enforced by GreenArrow’s configuration, but you will hit operating systems limits at some point.