I m currently using S3 for a project and it s working extremely well.

EBS means you need to manage a volume + machines to attach it to. You need to add space as it s filling up and perform backups (not saying you shouldn t back up your S3 data, just that it s not as critical).

It also makes it harder to scale: when you want to add additional machines, you either need to pull off the images to a separate machine or clone the images across all. This also means you re adding a bottleneck: you ll have to manage your own upload process that will either upload to all machines or have a single machine managing it.

I recommend S3: it s set and forget. Any number of machines can be performing uploads in parallel and you don t really need to notify other machines about the upload.

In addition, you can use Amazon Cloudfront as a cheap CDN in front of the images instead of directly downloading from S3.

The price comparison is not quite right: S3 charges are $0.14 per GB USED, whereas EBS charges are $0.10 per GB PROVISIONED (the size of your EBS volume), whether you use it or not. As a result, S3 may or may not be cheaper than EBS.

I have architected solutions on AWS for Stock photography sites which stores millions of images spanning TB s of data, I would like to share some of the best practice in AWS for your requirement:

P1) Store the Original Image file in S3 Standard option

P2) Store the reproducible images like thumbs etc in the S3 Reduced Redundancy option (RRS) to save costs

P3) Meta data about images including the S3 URL can be stored in Amazon RDS or Amazon DynamoDB depending upon the query complexity. Query the entries from Amazon RDS. If your query is complex it is also common practice to Store the meta data in Amazon CloudSearch or Apache Solr.

P4) Deliver your thumbs to users with low latency using Amazon CloudFront.

P5) Queue your image conversion either thru SQS or RabbitMQ on Amazon EC2

P6) If you are planning to use EBS, then they are not scalable with your EC2. So ideally you can use GlusterFS as your common storage pool for all your images. Multiple Amazon EC2 in Auto Scaled mode can still connect to it and access/write images.

You already outlined the advantages and disadvantages of both.

If you are planning to store terabytes of images, with storage requirements increasing day after day, S3 will probably be your best bet as it is built especially for these kinds of situations. You get unlimited storage space, without having to worry about sharding your data over many EBS volumes.

The recurrent cost of S3 is that it comes 50% more expensive than EBS. You will also have to learn the API and implement it in your application, but that is a one-off expense which I think you should be able to absorb very quickly.

Do you expect the images to last indefinitely?

The Amazon EBS FAQ is pretty clear; the annual failure rate is not "essentially zero"; they quote 0.1% to 0.5%. It s better than the disk under your desk, but it would need some kind of backup.