AWS Summit 2018 – Our Three Key Learnings

Raj Senanayake – Head of Engineering @Simble

This year Simble attended the AWS Summit 2018 as a shiny new AWS Technology Partner. The event was held at the equally shiny new International Convention Centre at Darling Harbour.

Our R&D team members, architects, developers and product managers all had their own area of interest and focus, which was certainly satisfied by the large number of sessions, workshops and streams available. We also found the free coffees and cakes a mandatory requirement to keep our team fuelled and buzzing from session to session!

Once back in the office we discussed and these are our three key learnings from this fabulous event:

1. Artificial Intelligence (AI) and Machine Learning (ML)

This year, there was a much larger focus on AI and ML. AWS Sagemaker was featured in a number of sessions and Amazon has gone to great lengths to make the product easier to use and integrate. A much simpler interface through Jupyter Notebooks allows you to code and document side by side, load your data, build your models, train them quickly and cheaply, ensuring you only pay for the compute power you need, when you need it.

The stand out was by far the session titled “The Ubiquity of AI and Machine Learning in our Everyday Life” by Olivier Klein, Head of Emerging Technology for AWS.  In this session an ML model was trained to identify a blow up unicorn and a guy in a hotdog suit, which was done with remarkable accuracy in a packed theatre. The key requirement for those looking into this technology is the essential need for good quality data.

2. IoT and Greengrass
Being an early adopter of AWS IoT technology ourselves, we were interested in understanding what’s new in the AWS IoT bag of goodies this year.

At last year’s summit we saw the official introduction of the Greengrass capability, a platform deployable to IoT devices to process data locally and securely communicate back to the cloud. This has now been taken to the next level. Greengrass now supports the running of a light-weight version of the AWS server-less processing framework, AWS Lambda, which allows similar functions to also run on your IoT device. The communication between devices back to cloud are authenticated and encrypted, with all the backend functionality such as the IoT messaging, caching and syncing between Cloud and the devices still supported.

A great new feature was ML Inference, where the Machine Learning models that are built and trained in the cloud can be deployed to the devices so that predictions can be made at the edge of the IoT network. This means you can get your heavy Machine Learning processes done in the cloud, push those models to the IoT devices, then allow your IoT devices to make decisions or take actions even when the connections are intermittent. Essentially these smarter devices allow smarter decisions to be made in a secure fashion.

3. Running and Scaling your Cloud-hosted System

There were many sessions covering numerous facets of running your cloud. Clever ways to reduce cloud costs, migrating to cloud, scaling and developing in the cloud and more. The session streams and workshops covered product offerings and technologies such as:

  • EKS (Elastic Container Service for Kubernetes) – A managed Kubernetes service that facilitates the deployment of highly scalable applications using containers. In addition it integrates with other AWS services such as IAM, VPC, CloudWatch and PrivateLink.
  • Cloud9 – A new cloud based Developer Environment (IDE) that is accessible via any browser. Multiple users can connect from anywhere facilitating pair programming and allowing easy group debugging between remote ream members. Also greatly increases the speed of development and debugging of Serverless functions both locally and remotely.
  • CodeStar – Streamlines creation and management of deployment pipelines to allow users to quickly develop, build and deploy applications (Currently only available in certain US Zones).
  • Xray – Traces, analyses, and visualises data from distributed systems. This allows us to review Request Behaviour, Application Issues and Application Performance.
  • Glue – Streamlines your ETL (Extract, Transform, Load) process via a fully managed service. It’s reusable, customisable, and cost effective due to its use of serverless technology to catalogue your data and make it immediately searchable and usable for analytics.
  • Kinesis – Processes large streams of raw data that can then be processed by other applications. Realtime data can be analysed instantly, pushed to S3 for long term storage, or sent to any number of other systems in your pipeline.

With a rapidly growing number of products launched each year by AWS (approx 290 in 2014 compared to over 1300 in the 2017/18 timeframe), the 3 day event is merely enough to give you a glimpse of that proverbial tip of the iceberg.

For companies like Simble who are already using much of the AWS technology stack, it is an opportunity to validate design decisions, make more informed predictions about technology shifts and most importantly to find the best solutions to meet our customers’ needs. As AWS put it, ‘It’s always Day 1 at Amazon!’, a mantra they live by and which has quite obviously helped them get to the scale and speed of operation they have achieved.

Bring on next year’s summit!