Setting Up a Tiered AWS Web Application Architecture in using AutoScaling and ELB within VPC
Sample AWS Web Application Architecture using AutoScaling and ELB within VPC
The purpose of this guide is to help you set up some of the AWS Web Application Architecture using AutoScaling and ELB within VPC.
This build comes from this AWS architecture document (except I am only deploying in a single Availability Zone (AZ), but modifying this to support multiple AZ’s isn’t too much more difficult and I’ll provide the parameters that need changing at the end). I’m only setting up 2 tiers in this example, but you can use the same concepts and commands to easily add a third tier and/or additional redundancy (ie: AZ’s) as you require.
At the same time, this guide will also show you how to deploy the right-side of the following image (Availability Zone 1B) taken from Deploying Elastic Load Balancing in Amazon VPC:
You can use the similar concepts and commands to create an environment on the left (Availability Zone 1A) if you would like.
What we’ll be setting up
- Everything will reside in the VPC.
- An Internet-facing ELB will load balance web requests across your Web tier.
- An internal load balancer will load balance the Web tier across the DB tier.
- Backend Web and DB instances are hidden from the Internet and reside in a private subnet.
- AutoScaling will help the Web and DB tiers remain elastic and scale based on load
- Elastic Load Balancing will load balance requests
- Private subnet
- Public ELB SG
- Private ELB SG
- Web SG
- Database SG
For AMIs, I am using the Bitnami LAMP images (they’re pretty awesome if you want to get up and running quickly with a LAMP stack and are new to or don’t want to set up LAMP yourself!):
- Web Server AMI:
- Database AMI:
Just give me the commands already!
Create ELB in public subnet in between web clients and the Web tier:
$ elb-create-lb publicvpcelb --scheme "internet-facing" --security-groups sg-06031c6a --listener "lb-port=80, instance-port=80, protocol=http, instance-protocol=http" --subnets subnet-65c6780c` DNS_NAME publicvpcelb-705800467.us-west-2.elb.amazonaws.com
Create launch configuration for Web tier:
$ as-create-launch-config testweblc --image-id ami-1821ab28 --instance-type t1.micro --key chanjoe --group sg-0b031c67 OK-Created launch config
Create AutoScaling group for Web tier:
$ as-create-auto-scaling-group testwebasg --launch-configuration testweblc --availability-zones "us-west-2a" --min-size 0 --max-size 1 --desired-capacity 1 --vpc-zone-identifier "subnet-66c6780f" --load-balancers privatevpcelb
Create ELB in private subnet between Web tier and DB tier:
$ elb-create-lb privatevpcelb --scheme "internal" --security-groups sg-da031cb6 --listener "lb-port=3306, instance-port=3306, protocol=tcp, instance-protocol=tcp" --subnets subnet-66c6780f DNS_NAME internal-privatevpcelb-65013364.us-west-2.elb.amazonaws.com
Create launch configuration for DB tier:
$ as-create-launch-config testdblc --image-id ami-1821ab28 --instance-type t1.micro --key chanjoe --group sg-05031c69 OK-Created launch config
Create AutoScaling group for DB tier:
$ as-create-auto-scaling-group testdbasg --launch-configuration testdblc --availability-zones "us-west-2a" --min-size 0 --max-size 1 --desired-capacity 1 --vpc-zone-identifier "subnet-66c6780f" --load-balancers privatevpcelb
A Common Issue
I get an error when browsing to my ELB:
HTTP Error 503 (Service Unavailable): The server is currently unable to handle the request. This code indicates this is a temporary condition, and the server will be up again after a delay.
HTTP/1.1 503 Service Unavailable: Back-end server is at capacity via *curl:
$ curl -v publicvpcelb-705800467.us-west-2.elb.amazonaws.com * About to connect() to publicvpcelb-705800467.us-west-2.elb.amazonaws.com port 80 (#0) * Trying 126.96.36.199... * connected * Connected to publicvpcelb-705800467.us-west-2.elb.amazonaws.com (188.8.131.52) port 80 (#0) > GET / HTTP/1.1 > User-Agent: curl/7.24.0 (x86_64-redhat-linux-gnu) libcurl/7.24.0 NSS/184.108.40.206 zlib/1.2.5 libidn/1.18 libssh2/1.2.2 > Host: publicvpcelb-705800467.us-west-2.elb.amazonaws.com > Accept: */* > < HTTP/1.1 503 Service Unavailable: Back-end server is at capacity < Content-Length: 0 < Connection: keep-alive < * Connection #0 to host publicvpcelb-705800467.us-west-2.elb.amazonaws.com left intact * Closing connection #0
- Check your healthy host counts for the ELB. HTTP 5XX errors can mean that you have no healthy backend instances to serve the request.
- Remember to open up the security groups to allow traffic from the Internet to the Internet-facing-ELB, from your ELB to the Web tier, from the Web tier to the internal-ELB.
- For troubleshooting connectivity inside your private subnet, launch an instance into your public subnet that you can reach and confirm you can reach the private instances and ELB.
Adding additional Availability Zones
- As I mentioned at the beginning, I did not deploy additional instances in other availability zones. You can do this if you wish by creating additional subnets (1 for each AZ you’d like to instances in) and adding them to the
--availability-zonesparameter in the
This is certainly not an all-inclusive guide and none of it is new information, but I hope it helps people from having to piece it together from the documentation below! Please let me know if you have any comments or suggestions.
Some helpful references
blog comments powered by Disqus