Menu
laidiferruni
By:
laidiferruni

Sending speed with Amazon SES is incredibly slow.

February 29, 2016 2.7k views
Email Messaging Server Optimization API

Hello guys,

I'm having a problem with one of my droplets. I have migrated my email marketing script from google cloud here in Digital Ocean and that is integrated with Amazon SES service.

Through my Google cloud instance and with linode (with similar parameters with the current droplets), I was able to send 50,000 emails (my current list) within 1 hour.

With this droplet, which is about the same performance as the other ones I had before, it takes more than 22 hours to send to my full list.

My current SES settings support a load of 90 emails / second, but the droplet is not allowing this to happen for some reason.

I tried upgrading ram and processor, but it didn't do any good.

When I see the graphs it says less than 1% of processor used at it's max, so I doubt it has anything to do with the instance in itself.

Maybe I'm just missing some server configuration or something.

Would really appreciate your help.

Thanks,
Laidi

Log In to Comment
3 Answers
ryanpq MOD February 29, 2016

That is an odd situation. We don't do any kind of rate limiting on droplets that would cause this type of issue so here is what I would recommend:

1.) Check out this doc at Amazon which covers a couple ways that sending via SES could be slowed down.

2.) Open a ticket with our support team and include the output of a ping and traceroute (or MTR) between your droplet and the SES endpoint (in case of any routing issue).

Reply
admin7f7f836bdb April 27, 2016

We have a similar issue using Python and SMTPlib.

What have you done to fix it? Sometimes SES is responsing very slow.

Reply
admin87f7492443 November 15, 2016

Hi,

I think there are connectivity issues between digital ocean and amazon datacenter. We have found those issues in AMS3 (where we have almost all our servers). Both SMTP and SES api is slow.

Also, when we switched to Sparkpost, their delivery of webhooks fails because of connection timeouts (sparkpost is also hosted on Amazon). We fixed this by addind a server in NewYork (digital ocean) who receives the webhooks and forwards them to AMS3 droplets.

We've did lots of tests and we are 99% sure that's the reason.

What datacenter are you in?

Reply
Have another answer? Share your knowledge.

Related Questions