THE AMAZON AWS SUMMIT LONDON, 2019

I attended the AWS Summit for the second time last week (Wednesday, 8th May). Here are my notes!
​
​I would estimate that it was at least 3-5x larger than last years’ event, based on the number and size of the “theatres” used as breakout rooms and the mini Wembley stadium that was the main auditorium!

As a mentor and provider of workshops and a programme for tech startups, I want to be able to point entrepreneurs in the right direction and to help them to understand their options from idea stage right through to planning, prototyping, build, launch and beyond. AWS is THE major player in terms of the number of cloud service offerings available, offering access to many of their services for free to startups for 12 months via their “AWS free tier” initially, and at non-prohibitive price points thereafter, based on pay-as-you-go usage rates.
The number of services AWS provides is evolving at an impressive rate, so I attended the Summit to ask questions, and get a handle on new services.

It was tough to choose which breakout sessions to attend, (please see the images below) but I chose ones covering machine learning (Become a machine learning developer), computing storage (Amazon EC2 foundations and cost optimisation), microservices (modern application architectures), systems architecture (Are you well architected?), and the use of AI services in applications (Build intelligent apps using AI services).

A huge choice of sessions to attend at Amazon's AWS Summit, 2019

​Summit attendees were free to engage in a little "lightning learning" at set times just by sitting in front of the screen and putting on the head phones provided...

Breakout sessions
A book could be written to cover all the AWS Services and their applications (and I have promised myself not to write any more for now!) but I have summarised some of the sessions I attended and some of the use cases presented, along with links for further investigation.

Session 1 - Become a machine learning developer
Presentation from Lars Hoogweg, CTO, Lebara.
Lebara presented their own case study, discussing the different categories of fraud that the business faces on a regular basis. Lars explained that as a result of using SageMaker, (AWS’s machine learning service) and bringing together subject matter experts from AWS and Lebara, this had enabled them to detect new cases of fraud AND identify further opportunities for fraud detection within 3 days of using the tool.
They are currently planning to put these learnings into production.

A huge audience at the machine learning session, and this wasn't even the largest conference room!

Amazon SageMaker is part of AWS’s 12 month free tier for startups: https://aws.amazon.com/free.

https://docs.aws.amazon.com/sagemaker/latest/dg/how-it-works-mlconcepts.html

Setting up your machine learning model. Copyright, Amazon, AWS.

​Session 2 - Amazon EC2 (Amazon Elastic Compute Cloud) – Computing storage at scale
AWS offer pay-as-you-go services, but for larger, and very resource intensive activities, this payment model can become expensive.

This is less of an issue for early stage startups, but it’s important for all businesses to forward plan, and I strongly recommend that new startups consider whether the free, or low-cost services they are using today will still be a viable option as they grow. Usage levels increase over time, which can push businesses onto higher price tiers. It’s important to do at least a few sums and some due diligence based on projected future use to see if the service will still be affordable in the future – if not, it means uprooting and moving away from a service you may be otherwise satisfied with to search for alternative suppliers, which can take time, and tie up internal resources.

AWS actively supports clients in understanding the value of rightsizing – optimising usage and performance for the most economical price, and offers autoscaling options to achieve a balance. The service assists in a number of ways including scaling dynamically which adjusts services based on “targets” set, and predictive scaling based on trends, and spikes in traffic. The scheduled scaling service can be used to scale and cope with load, adjusting at the right time(s) based on usage trends which are known and already understood by your business.

Amazon also offers various instance purchasing options, including Spot – effectively a borrowed, but cheaper resource which Amazon can take back with 2 seconds’ notice, but which is useful for flexible, fault tolerant services running within a business, On demand – useful for handling spikes in traffic, and Reserved Instance (RI) – for steady, ongoing use.

​
Session 5 - Build intelligent apps using AI services
I went 50/50 on the 4th and 5th sessions of the day, and moved on from the Are you well architected? session, and luckily managed to catch the live demonstration and case study delivered by Gareth Butler, Senior Programme Manager, at HSBC during this session.

With over 235,000 employees, HSBC have a heavy requirement from internal staff for information. The number of queries is both large, and repetitive in nature.

HSBC have used AI to create their HSBC Risk Innovation Assistant - a text and voice based information service for staff.

Results are returned with matches highlighted in text, and the assistant is also able to “reply” with answers to employees.

The assistant is multi-lingual – it detects the language of the query and returns results in this language. (Amazon Translateoffers more than 12 languages including 2 Chinese languages, Arabic, Turkish, French, German, Portuguese, Spanish, Russian and Japanese.)

We saw a live demo of the assistant – Gareth asked it a compliance related question from the stage. The system then replied back to explain the nuances of the relevant compliance rules, and highlighted the potential exposure of HSBC to reputational risk.

To set up this sort of service, documents first need to be ingested into Amazon S3 (Amazon Simple Storage Service), where the content is tagged. It must then be passed to Amazon EC2, which then allows the data to be utilised by Elastic search.
​
The “conversational” part of the service is made possible thanks to Amazon LEX which handles automatic speech recognition, ASR, and its understanding of natural language makes it possible for LEX to understand the nuances of speech. The demo was pretty impressive from this perspective - Gareth’s question was answered with the briefest of pauses and LEX understood the context of his question, providing a direct answer to the question asked.

Other observations
The large number of data warehousing companies presenting at the Summit. I didn’t count them, or compare the relative ratios of DW companies to other types of business (sorry, I'm being a nerd here, but that’s how you’d prove this objectively, right?) However, I don't recall seeing as many of these types of companies in attendance last year.
​
Does this reflect a growing trend as more businesses recognise the importance of making data driven decisions, and collect increasingly large data sets in order gather more insights? In addition to data warehouses, businesses are now also using data lakes, whereby data is pooled ready to be prepared and utilised for different purposes when needed. (The link provided goes to a site which offers a neat description of the differences between data warehouses and data lakes.)

Some more images!
​The innovation flywheel
Idea - experiment - feedback - repeat!
​
This was taken in the main auditorium - the "mini Wembley stadium" I mentioned earlier - look at the size of the screen relative to the speaker - it was monstrous!

The main conference area and exhibition stands

​Characteristics of a well architected system

This year there was a nice startup section helping startups with the setup of their basic systems architecture...

Did you attend the AWS conference? What was your impression?