Reflections on AWS re:invent 2016
I am just back from Las Vegas, a trip I gratefully made with 3 other colleagues on behalf of @KainosSoftware. I have been to many conferences before but this was my first time at re:invent. It is an understatement to say I was blown away by the scale of it (a remark that stands tall in the context of the AWS platform also). In only its 5th year over 30 thousand people descended onto the strip for 5 days of learning, networking and good times.
A range of talks and activity were spread out across the week. Monday was focused on industry break-away sessions and hackathons. I particularly enjoyed the Public Sector breakaway session where I got to spend time with a number of peers working in public sector across EMEA, but I certainly didn’t enjoy the crowds or the queue’s, which were quite horrendous. Tuesday followed a similar pattern to Monday with particular highlights including sessions on security testing and threat modelling, but more about that in a separate post. Wednesday was the first great day with @ajassy’s keynote kicking things off. This was a fantastic talk with some great stories from many AWS customers. Personally, I would liked to have seen this keynote right at the start of the week to really kick things off and allow all the announcements to be better socialised among the delegates throughout the remainder of the week. On Thursday @Werner presented the technology keynote. This was a genuine highlight and if you weren’t at re:invent and you only have time to watch one recording, make it this one. But again, I would have loved to have seen this at the start of the week. People were inspired after this talk and everyone was talking about it. Why wait to Thursday for it? Monday or Tuesday next year please, Amazon.
It was a BIG conference with a LOT of content so trying to summarise the themes is difficult but there a number of areas that I felt came across strongly.
- Not Iaas — If you still think of AWS as just IaaS you are wrong. Yes, IaaS is their bread and butter and you can still leverage these building blocks, but it is no longer transformational. The transformations will come with the applications built on the rich set of higher tier services they provide around analytics, machine learning, IOT and media recognition. The obvious question/concern that comes with this approach is vendor lock-in, but Amazon are innovating at such speed and scale my view is that if you want to take advantage of this you need to accept an element of risk here and architect your applications accordingly. There are rich import and export services if you ever get to the point of needing to move your workloads somewhere else.
- Serverless — This was a big and recurring theme of the week. A number of new features where announced around Lambda and it is clear serverless is going to play a huge part of Amazons (and Google’s and Azure’s) future. Say it quietly, but is the writing on the wall for your WebOps team? I’ll not answer that just now, but it is clear there are many advantages with serverless and if it delivers on its promise and gains wide scale adoption we will see a world where we worry less about infrastructure and more about the application, the UX and how to get to market faster. Sounds like a plan to me. You will pay for exactly what you use, not for EC2 instances that are only partially utilised. The cost savings could be significant.
- IOT — a huge focus in this area and the associated machine learning and analytics services. I think AWS Green Grass in particular will gain a lot of traction and hopefully stabilise IOT development, providing a platform for developers to build hardware devices quickly and more securely, again leveraging Lambda.
- Modern Data Architecture — Until recently Amazon had a number of good, but fragmented services around data processing and storage but they have doubled down in this area to provide a complete end to end data platform, covering the entire lifecycle of your data, from ingestion though to Analytics using tools like Glue, QuickSight and Athena. Buyer beware: don’t get lost in the buzz word bingo — make sure you pick the right tool for the job.
- Image and Voice — The current state of the union on voice recognition is heavily focused on intelligent personal assistants. I don’t resonate personally with this, but I’m probably just getting old. I don’t talk to my devices and I don’t get along very well with Siri, but the next generation are different. I see it even now with my 6 year old son who finds it completely natural to talk to the TV or the iPad. But Amazon are looking beyond this with services like Polly and Lex for voice processing and Rekognition for Image recognition and analysis. Lex in particular looks very interesting, using automatic speech recognition and natural language understanding to expose conversational interfaces that understand the context and the intent, something badly missing from services like Siri. Imagine use cases like a patient being able to book an appointment to see their GP or paramedics interacting with a bot to get and interact with a patients medical history. So, if you are not seeing voice and image recognition now, you will in the future and Amazon will be leading the pack with services like Polly, Lex and Rekognition.
There were such a huge number of services launched we will need to divide and conquer within our teams in @KainosSoftware to understand the technology and use cases better. I’m looking forward to learning more about some of the services that will add value to @KainosEvolve. Things like PostgreSQL for Aurora, Glue, X-Ray and Athena all show a-lot of promise but require further investigation to go beyond the marketing blurb and understand more.
Overall, a superb conference and a great experience. Thanks for sending me, @KainosSoftware