Travelex at AWS re:Invent
AWS re:Invent is one of the biggest cloud technology conferences of the year. This time around, there were around 50,000 attendees, including a lucky handful from the Travelex team. What follows is their key insights and analysis from the event.
Containers are First Class Citizens
It is no secret that our engineering team is passionate about the use of container technology, which underpins Travelex Wire. We use containers to accelerate release management and improve the resilience of our services, whilst keeping the engineering function lean. As well as presenting on compliance in a container-based environment at re:Invent [LINK], members of our team also participated in focus groups to help shape the future roadmap for Amazon’s container offerings -- the announcements of a managed Kubernetes service and a fully-managed container environment, FarGate, came as no surprise.
We are looking to a future where, instead of the servers of old, we have units of compute power sized into “big” (a.k.a. virtual machines), “small” (containers), and “tiny” (serverless functions). At this point it’s clear that the goal for each of these categories of compute is to have identical security, availability, networking and monitoring characteristics, giving software engineers the ability to minimise the amount of undifferentiated heavy lifting required to run their workloads. We’re definitely still on the journey, but there are already the beginnings of a better formula for new services.
Machine Learning and DeepLens
One of the big keynote themes was how to improve the human interface into systems and services, primarily with voice and video, supported by machine learning. There is a clear leadership battle in this arena, with AWS seeking to differentiate from the likes of Google and Apple with video input. Despite hundreds of people queuing, two of our team squeaked into the DeepLens workshop and came home with a pair of these devices: essentially eyes and ears for a cloud-based brain. We’re very excited by some of the potential applications of this technology, such as KYC on approach to a store window or real-time sentiment analysis during a transaction. The next hackday should be very interesting.