1

I am running a Drupal website that is using RDS for the database. I have 2 instances, maybe more in the future, that are load-balanced. For 60-75% of the year the static content will not change. The other 25-40% of the year we will be developing modules and functionality that will require file modification several times each day. I would like to implement a system that will handle file sync between n instances. This system should be as hands and thought free as possible. Note: These instances are only web servers with ssmtp, (yes it is ssmtp not smtp), for php mail - nothing more.

Some people state that one can use S3 to host the static files and serve them through EC2 Instances. Others use free tools like Kwatee or paid ones like Scalr. I even saw some people using EBS Images and duplicating them for mount to the EC2 Instances - this seemed like the worst method. The last method I saw used rsync to accomplish the file distribution.

How can I properly compare these methods and come to conclusion about which implementation to choose? Also, I do not fully understand the implementation of the S3 method. Because I know it will come up, cost is of course a point of interest, but ease of use and reliability are more important at this point. (Within reason of course.)

stackoverflow References:

Load balancing web servers + keeping content synced

Deploy Content to Multiple Servers (EC2)

Where to put shared files on amazon AWS design for failure architecture?

Community
  • 1
  • 1
Brandon Bearden
  • 820
  • 11
  • 29

2 Answers2

2

I would recommend to keep your static files as S3 objects. Amazon S3 is a highly durable and reliable storage of binary objects (files), accessible in public web through HTTP/S (if desired). Besides that, you can make these objects accessible by your clients through AWS CloudFront (content delivery network).

yegor256
  • 102,010
  • 123
  • 446
  • 597
0

Point 1) Drupal works well with GlusterFS. You can create one or more GlusterFS EC2 instance as a common storage pool. You can attach all your Drupal EC2 instances into the common GlusterFS Storage Pool. Whatever data , files you put on the common storage pool is readily accessible across all Drupal EC2 instance immediately. This is general technique used in scalable Drupal deployments in AWS. This model avoids rsync, S3 and other code changes in your architecture

Point 2) If you have option to change code then integrate it with S3.

Harish Ganesan
  • 743
  • 5
  • 4