AWS:reInvent Conference Report

AWS:reInvent Conference Report

AWS:reInvent is the annual Amazon Web Services conference, and this year two MSI employees, Jon and Dan, were able to attend and brought back some interesting information.

It is quite possibly the best source of information for anyone that even tangentially uses AWS, or has the desire to learn AWS. It is overwhelming – 32,000 people spread over two full casino conference areas – and information dense. It is a very long week. That said, there really is no better way to learn about the services offered, and how to properly use them. Nor is there a conference that gives you more for your dollar, whether you like swag, networking, or general social events.

There are talks to educate attendees on every phase of using AWS. These include planning a migration to the cloud, improving deployments, securing services, “microservice-ing,” and more targeted deep dives into specific features of particular services. This year also had a focus on how to use and develop echo/Alexa skills in a work environment. There are workshops, training for certifications, certification tests, structured networking opportunities, a colossal expo with hundreds(?) of vendors, and plenty of after hours events.

The biggest issues with the conference are its size and overall denseness of information. There were lines for absolutely every talk, bathroom, refreshment stand, etc. The information density results in a lot of duplication between talks, as well as buzzword fatigue. By the end of the week you are pretty well burnt out on anything “cloud.” Many talks can be hard to follow because the information is too fine-grained.

Here is a quick overview of some of the more notable take-aways:

  • Some new services that are particularly fun or interesting:
    • Postgres on Aurora
    • Athena — Query S3 with SQL
    • Lightsail — Launching a full VPS from nothing with something like 3 clicks
    • Snowmobile — like Snowball, but an actual 18 wheeler driven to your datacenter. They drove the semi onto the keynote stage (nice touch). It is for customers with exabytes of data — using conventional methods would take something like 26 years, using Snowmobile and 10 semis you can cut that down to approximately 6 months.
    • All sorts of new EC2 instance types for Big Data stuff, or GPU intensive stuff, or CPU intensive stuff, etc.
  • Here are a couple talk highlights:
    • Capital One’s Alexa skill is incredible, and we can learn a lot from how much work they put into making it as natural as possible to interact
    • Cloud Watch events and Cloud Trail are INCREDIBLE for all sorts of things, not the least of which is monitoring users/developers for security concerns
    • Polyglot nature is a huge benefit to microservices, use the language best suited for the task
    • Lambda is unbelievably cheap; one speaker mentioned 3 million requests for $18 a month
    • A great strategy for moving from monolith to microservice is to “strangle the monolith”
      • Start on the fringes and extract the small, easy things. Continue moving towards the hard stuff in the middle.
      • Eventually you’ll have several smaller monoliths, so you rinse and repeat
      • Make sure you keep business needs in mind as you strangle.
    • The JPL gave a great talk on how they use Echos to automate their office
      • Clean rooms are now voice based
      • Conference room registration is voice based
      • Monitor, have risk tables that map to actions, take action, THEN notify people about what happened and what was done.
    • aws-cli now has aliases
      • can be full on shell functions which is amazing for scripting service creation
    • Resources if you’re interested in any of this:

Finally, this conference is a great time. It can be hard to hold onto your sanity but there is a lot of great information, and an amazing conference-end party.