We make extensive use of AWS infrastructure @MyDrive. We’ve been using them for a long time, initially using them to mirror a more traditional infrastructure using cloud resources, to allow portability between cloud providers.
More recently we’ve identified that we can make considerable cost and time savings if we start using more of the baked in AWS Managed Services and we’ve begun that transition, for example we have already moved our realational database infrastructure to RDS.
We’ve also been cross skilling Developers and Operations staff to allow us to cover leave and swarm more effectively on tasks. My background is as a Developer so I was looking forward to learn more about what services we could use to allow us to deliver faster, and more consistency. I’m also on our Pager Duty rota, so I’m also obviously very interested in services that make the operational maintenance someone elses problem!
With this in mind we took most of the Software Engineering team to AWS’s free AWSome day, which is a one day event they hold a few times a year in London, and are taking on tour around Europe, so we could learn what services we could be taking advantage of already, and what was available to look to use in future.
The day started fairly slowly for us, being a team that are already actively using AWS in production most of the introductory morning sessions were covering things we were already familiar with on the computational and networking layers. It also felt a bit sales heavy, which is understandable, given it was a free event and many of the attendees were new to AWS.
The afternoon sessions were much more useful and got into the meet of what we were interested in, focusing on many of the managed services we don’t use and there are a few services I think we can use, and a few potential projects for the next Hack Day.
Archiving old data to Glacier should save us a bit amount on storage costs. Although the savings are small, it’s so easy to set up automated archiving rules on a per bucket basis that there’s almost no cost to implement archiving.
I hadn’t looked much at DynamoDB but it does seem like it might be a good addition to our database infrastructure. It’s a simple No-SQL key, value pair store. With each object limited to 400KB. One nice feature is that you can configure your required IOPS so it’s trivial to scale. It’s also comfortable with updates, and supports secondary indexes. There are certainly some small, high throughput, simple data sets where DynamoDB sounds like it would be a better fit than Cassandra or a Relational Database.
We’re already looking at replacing Redis with ElastiCache, but I think we can also take advantage of it to cache some common requests.
They didn’t go into OpsWorks in much detail on the day but from my initial research it looks like We might/could be able to take advantage of it to replace hosted Chef in some areas.
On a personal note, I’m going to register for a personal AWS account and move a bunch of old static websites to AWS ), for a greatly reduced cost (free for a year).
I’ve also registered for the AWS Summit which looks to have some good hands on technical sessions.