AWS Load Balancer is slow?

Question

I realize I should probably ask this question on AWS forums but I tend to get quicker responses here.

I'm trying to setup a load balanced environment in AWS. I have two EC2 instances that are mirrored and both are accessible via their own public IP address. I have the load balancer set up with the two EC2 instances added to the balancer and it's reporting that the instances are healthy.

What I'm finding is that when accessing the Load Balancer directly like the url they gave me, xxxxxxxxxx.us-east-1.elb.amazonaws.com, page loads are ridiculously slow... like 45-70 seconds long per page load. When I access the public IP directly to one of the instances we're talking like 3-5 seconds page load.

Any ideas as to why this is so slow?

Solution

The reason for the delay is because you have the ELB setup for multi-az without any application instances in the other 2 AZ's configured. Without instances in those AZ's requests will tend to fail because the ELb still returns IP addresses for those AZ's even if there are no active application instances. Please disable the other AZ's for now and continue your tests.

Comments

[Admin]

Is the ELB forwarding the port traffic as you'd expect? I.e. you're passing port 80 to port 80, listening on 80 on the instances, and not doing anything complex or unexpected with the routing?

[Admin]

I should not be slow. Amazon created a paper describing the best practices for ELB. I suggest you taking a look at it. aws.amazon.com/articles/1636185810492479 HTH

[Admin]

I am having the same problem, ELB works fine if there is one EC2 instance behind, after I added one more instance, sometimes the connections just hangs and resulted page failing to load

[Admin]

Are you using an external or internal ELB? Have you tried setting up an external ELB and then putting your instances behind that ELB? In this setup your DNS will access the external ELB and your external ELB will load balance to your non-EIP EC2 instances. You can also monitor performance using CloudWatch as well as enable Access Logs on your load balancer to help troubleshoot. I have run into this issue before but the problem was on the application side not being able to work with load balancing.