AWS Summit 2019

8 min readJun 3, 2019

The AWS Summit brings together the cloud computing community to connect, collaborate and learn about AWS. This year’s summit proved to be the largest one-day summit on the planet for AWS with over 12.000 attendees. The event took place at Excel London, offering an Expo center with AWS experts, customers, and partners, tens of breakout sessions, waffle serving, gourmet coffee, and an escape room!

The most discussed topic this year was AWS DeepRacer, a 1/18th-scale race car driven by reinforcement learning. Aiming to give a fun start with machine learning techniques, DeepRacer is the first global autonomous racing league. More details about the competition can be found here.

Keynote

The event started with a keynote by Matt Garman, VP of AWS Compute Services, who was the first AWS product manager. The keynote was an overview of the AWS current infrastructure and services as well as announcing some new additions, amongst which: 4 new AWS regions and a new EC2 I3en storage-optimized instance for data-intensive workloads. Matt also talked about the growth of AWS, having over 100.000 customers in the UK and a 41% growth in 2019 Q1 compared to 2018, reaching almost $31B.

A great example of a UK customer is Sainsbury’s, a 150 years old business which two years ago was not in a happy place, struggling with digital transformation. After adopting AWS services, Sainsbury’s moved from 5–6 releases per year to releasing multiple times per day. “Our relationship with AWS really kicked off at the point we decided to take our groceries online business and rebuild it in the cloud. This was effectively taking a WebSphere e-commerce monolith with an Oracle RAC database, and moving it, and modularising it, and putting it into AWS,” Sainsbury’s CIO Phil Jordan told the audience. In this blog Rowena Parsons, Senior Solution Architect at Sainsbury’s is explaining how they use some of the AWS services, including AWS Lambda.

MoJ, The Ministry of Justice, has defined a vision that all systems operating on UK official data classification will be migrated to the cloud by 2020. Tom Read, Chief Digital & Information Officer, explained how MoJ is such a unique place to work in and how the reason for innovation is not the competitive advantage. Being a monopoly service that people use in very vulnerable moments of their lives, the organization’s mission is driven by this responsibility to deliver their services. Similar to Sainsbury’s, MoJ was dealing with frequent service outages, an old on-premise datacentre, and deploying only a few times per year which directly impacted the accessibility of services for citizens. Once the Government Digital Services, GDS, imposed the use of cloud, they started containerizing and immediately saw great improvements in reliability, savings, patching, and monitoring. Looking ahead, the plan is to move to a Kubernetes stack, cloud-native platform and break everything into microservices.

More examples of AWS customers and their achievements:

Stagecoach migrated its revenue reconciliation system to VMware Cloud (AWS Partner) in just 3 weeks;
Comic Relief uses AWS Lambda as its core donation platform taking 300 donations per second, and reducing the IT infrastructure spend by 93%;
McDonald’s launched a home delivery program based on microservices using ECS in 4 months, scaling up to 20.000 orders per second with less than 100ms latency. That’s a lot of hamburgers!
Photobox brings over 20 million customers moments to live with their photos; they migrated their core e-commerce platform photo database to Amazon Aurora improving performance by 30% and savings with more than £450.000 pounds per year. Photobox also moved its data storage to S3, which is 5.7 billion photos or 9TB of data, saving £4.5 million per year.

DevOps at Amazon

Lastly, Amazon broke its monolith architecture into microservices 15 years ago which brings me to the first talk I attended, DevOps at Amazon, given by Daniele Stroppa, Solution Architect. Daniele explained how Amazon split their workforce into groups called 2 pizza teams, which formed a DevOps team with full ownership of the project, tech agnostic, and self-service, not depending on other teams. The beginning was good, small teams with small projects where best practices are encouraged building single-purpose services, however, there was still only one pipeline. This was another required change, having multiple pipelines. Amazon easily reached thousands of 2 pizza teams, making use of microservices, CI/CD, multiple environments, infrastructure as code, and using several types of deployment. The talk ended with some food for thought, what if 2 pizzas are 1 too many? “If fewer people are needed to build things, can we build more things & faster?” was the question which Daniele left us with.

Building modern APIs with GraphQL

The next session was about GraphQL, an open-source API query language used by Facebook, Amazon, Microsoft, Twitter, Virgin Trains, and more, which can help tackle modern API design challenges. The session was delivered by Robert Zhu, Principal Technical Evangelist, who took part in the creation of GraphQL at Facebook. Along with a technical demo, two important questions were answered: why and when to use GraphQL?

Why? There is only one network call, all requests are sent to a single /graphql endpoint, following the reducing complexity through abstraction programming principle. Secondly, GraphQL gives you exactly the data that you ask for, solving REST’s efficiency problem of over-fetching and under-fetching.
When? If you have a RESTful reliable service that is stable and tested, it’s probably not worth throwing it away. With GraphQL you can implement certain design patterns on new or existing web services. One way of visualizing this is the following graph.

Despite its advantages of easily expanding, GraphQL can present a steep learning curve and if you find yourself above the design payoff line, it’s probably not worth introducing change and risk.

Modern applications architecture

Another breakout session was about developing modern applications in the cloud by Ian Massingham, Director, Technical Evangelism and a customer speaker Ann Ledwith, Director: Delivery Optimisation at Sage.

Invention requires two things: the ability to try a lot of experiments, and not having to live with the collateral damage of failed experiments.

With AWS CEO Andy Jassy’s words in mind, the talk focused around some key best practices when it comes to building applications: security and compliance, microservices, serverless technologies, infrastructure as code, CI/CD and monitoring. These map to the capabilities of a modern application: secure, resilient, elastic, modular, automated and interoperable.

Using AIOps to reduce incident volumes

Lastly, Paul Ferguson, Senior Consultant at AWS and Itai Njanji, Global Practice Manager delivered a talk on using AIOps tools to reduce noise from monitoring events and tickets and increase the service quality.

Step 1: Set up for observability. You can do this with CloudWatch to gain visibility on your cloud resources.
Step 2: Collect and store all logs and data. For this, there’s a variety of AWS services: Kinesis, S3, Redshift, Elasticsearch service, etc.
Step 3: Querying and patterns mining using Amazon Athena and SageMaker.
Step 4: Alerting. After the intelligence operates, notifications and remediation actions can be performed using Amazon SNS and Amazon Lambda.
Step 5: Integration with your ITSM system.

Other topics and facts:

AWS GetIT program (already piloted in the UK) to inspire, train and develop Year 8 young ladies in UK schools to consider technology careers and opportunities.
Amazon Aurora is the fastest-growing service in the history of AWS. Combines the simplicity and cost-effectiveness of open source databases with the performance and availability of commercial databases.
Today, AWS hosts more than twice as many Windows workloads than Microsoft.
The new AWS Cloud Map aims to keep track of all your application components, their locations, attributes, and health status.
AWS App Mesh provides an application level networking for all your resources; a single point of control for communication between services so you can easily pinpoint exactly what’s going wrong.
With IoT technologies thriving more and more, device data is becoming critical to achieving precise insights through ML, predicting 75 billion connected devices by 2025; examples of applications are remotely monitoring patient health, Alexa integration with Vizio devices, etc.

Put machine learning in the hands on every developer and data scientist; make it more approachable.

AWS wishes to achieve this with SageMaker. A customer example is British Airways which uses SageMaker to anticipate new patterns and suggest when aircraft may need minor or major maintenance.

HSBC converted bank policy docs into a conversational bot that reduces the time to answer a regulatory question using Amazon Translate, Comprehend, and Lex.
Marinus Analytics used Amazon Rekognition to build a system that finds human-trafficking, scanning photos of victims and comparing with photos online.
85% of all TensorFlow workloads in the cloud run on AWS; there comes an inference cost issue which AWS addresses with AWS Elastic Inference to add GPU acceleration; AWS Inferentia ML chip will be available later this year, and with EC2 G4 instances coming soon, AWS is promising more and more services designed for ML applications.
Amazon Forecast and Personalize are in the preview stage and both came from Amazon’s experience of using them. Like most AWS services, these are fully managed, you just need to point at your data to get your forecast or personalization.

From a placement software engineer point of view, and being quite new to the cloud community, you can easily get the “imposter feeling” and maybe decide against attending this kind of event thinking “I’m not ready yet” or “it’s too in-depth for me”. However, you have the opportunity to hear out lessons learned from customers and partners, gain insights into what is happening today and what’s possible to do next. At the end of the day, you feel extremely tired, but excited and motivated to see more. It’s incredible how events like this help you move forward, even just by being there, it’s impossible not to feel the excitement to go and build great things. It’s impossible not to start feeling like you’re one of them. Actively being part of a community, whatever it may be, can be very empowering.