Speed of sending - some experiments

jamescridland

There is some good advice in the FAQ about setup of servers for speed.

I've been sending a newsletter to about 15,000 people, and started doing this about three years ago. I only recently, dimly, realised that I've been sending to Amazon SES's East US servers from a bespoke AWS Lightsail box in Ireland. If you're interested, I can therefore report back that if you send from "the wrong" SES region, your mailing takes about 25% longer to send since I've been able to do that experiment. (If "experiment" means "get everything wrong"). I hope that data point is useful for some here.

My Lightsail box is 1 GB RAM, 1 vCPU, 40 GB SSD, and only used for Sendy. It only goes up to about 15% CPU use when sending. For resilience, I also use the not-recommended Amazon RDS. That jumps to about 45% CPU use during sending (and sends around 5.5MB/s of network traffic). This is clearly the bottleneck in sending - and I only mention it here because Ben's relatively clear that you shouldn't be using RDS, and this might back him up in that view! I do not, however, ever get any timeouts or issues with sending - it's been super reliable and very excellent.

I'm perfectly happy with Sendy's sending speed. It's fine for my use: but, since I've been reading other threads here about sending speed, thought I'd chime in with my experience in case it helps future users.

Ben

Thanks for sharing James.

Yes, using the correct Amazon SES region helps dramatically. This is covered in the 'sending speed FAQ' here → https://sendy.co/troubleshooting#sending-speed

And yes, users who use RDS experiences slower sending speeds and unforeseen issues. RDS is good for 'storing' data but not optimized for quick real time data processing like what Sendy needs.

Here's one feedback from a user using RDS:

Screenshot:  

Using a database hosted on the same instance that is hosting Sendy will speed up sending speed as well. This is also covered in the sending speed FAQ.

Thanks.

Best regards,
Ben