The 2nd day in Las Vegas at the AWS re:Invent started by one of the most important keynote. It was the keynote from Adam Selipsky, the new CEO of AWS replacing Andy Jassy who moved to Amazon after leading AWS since 2006. As usual, the keynote is a good opportunity to speak about the past successes and announce new services or products. Adam announced a new generation of the ARM processor built by AWS Graviton3 and new instance type using it. The new launches covers a broad range, from Redshift and Analytics being now available in serverless which can be interesting on our area to a new service to deploy private 5G on site.

After the keynote, the rush to learning continued and I attended different interesting sessions.

Chaos engineering

The name itself of the method may feel people a bit uncomfortable, I don’t know anyone interested in having chaos in its infrastructure. But chaos engineering does not mean breaking something randomly in Production and see what’s not working anymore.

I see it more as an extension as what we usually advice our customers: plan high availability and more important test it regularly. Chaos engineering may go deeper but in essence is in the same spirit, prepare a test and validate it works as expected when something fails.

In this builders’ session, we had the opportunity to try to stop instance in an auto-scaling group using AWS Fault Injection Simulator. This service allows you to create experiments and put some limits to stop the test for instance if the response time of the application is too high.

Moving data to AWS

There are 2 main ways to bring data into the cloud: online vs. offline. It seems a bit obvious but sending data to a public cloud provider will consume network bandwidth, by definition the cloud provider being outside of our premises. When choosing between online vs. offline strategy, the amount of data to transfer will be a key element as the available network bandwidth.

Online migration is done with AWS DataSync or AWS Storage Gateway services. It does not have to be only in 1 direction. Formula One is using the combination of both services to archive data in the cloud and then retrieve it to use it when needed. See more details at https://aws.amazon.com/blogs/storage/how-formula-1-uses-aws-datasync-and-aws-storage-gateway-for-backup-and-archiving/

Offline migration may be mandatory when the bandwidth is not large enough or on air-gapped environment. Here, the products to look at are Snowcone (having DataSync capabilities), Snowball or even Snowmobile (a full truck) depending on the amount of data.

AWS CLI tips and tricks

I’ve finished my day with a workshop about AWS CLI. I’m big user of CLI and I was already fan of –query option using JMESPatch but I still learned new tricks.

I’m now big fan of AWS CLI aliases. As we have multiple accounts and used temporary credentials, one of my top command was:

$ aws sts get-caller-identity

I’m happy now to be able to run instead:

$ aws who

Let’s see what I’ll learn tomorrow.