AWS re:invent re:cap

AWS re:invent re:cap is the free-to-attend event held by AWS in London 22nd January 2020 at the etc Venues in Houndsditch. The focus this time is on the Public Sector.

I registered for this event in December, prior to leaving my previous employer (Vaimo UK) and am attending as part of my new role at We Are Kitty (a TIPi Group company). My main interest is to explore more of the AWS tracks - how, why and main selling points vs self-hosting or other cloud providers (Google, Microsoft). Having used AWS under the guise of Magento Cloud (launching the Helly Hansen website in the first of its kind and size), I've been keen to explore how to build on top of this solution without restriction.

At the end of this piece, you will find my wrap-up of the day - what did I find useful, what do I want to follow-up on and what may not have been as useful.

Before reading further, please understand - the below is my interpretation of the day's events and wording. The words are my own (minus direct quotes) and not reprsentative of Amazon or my employer)

Today, I'll be following the Management Track and at the next one I will aim to review the technical track. You may come to realise - this is just me taking notes that I can then reference anywhere, long as I have internet access.

Arriving at the venue, there were no easy directions to say "go to the second floor", but  a polite staff member was repeating "go to the second floor". On arrival to the second floor, registering was easy with plenty of people available and then on to networking space with breakfast, AWS staff and partners (Cognizant, Softcat, Fujitsu etc).

Breakfast table at AWS re:invent re:cap at etc Venues

Introduction Keynote

On to the introduction keynote by Chris Hayman, Director UK and Ireland Public Sector.

  • 40,000 public sector clients
  • New services: Sagemaker Studio, Transcribe Medical, Outpost general availability
  • Regional launches: Connect, Cloud 9, Firewall, Lake, Translate, Dynomo DB
  • Inference (custom silicone, ML)
  • Graviton2 (price/performance for cloud workloads)
  • Amazon Braket (quantom computing)
  • Nitro Enclaves
  • Windows server: end of support migration and license management
  • AWS data exchange (find and subscribe to third party information)
  • Alexa Voice Service (AVS) for IoT core
  • AWS Outposts - AWS in on-prem environment (ECS, EMR, RDS, S3 (in 2020))
  • Local Zones - first instance will be in LA
  • Wavelength (5G edge location capabilities)
  • AI / ML - deepdive into Transcibe Medical. Embark is AWS's way to get you onboarded and trained with support to launch ML journey

Tim Hinchey, Head of Arch at DVSA

  • Machine Learning journey
  • Aggressive Cloud implementation; desire for zero data centres
  • 50 million transactions (main one is MOT service)
  • Challenges (private connectivity, data centre closures, hardware procurement (difficult in public sector).
  • AWS Connectivty Hub
  • Standup services in hours, instead of months.
  • FN4G (Future Network for Government - run by Digital Cabinet Office)

Back to Chris Hayman

  • Mission - fast, cheap, scale. Retire technical debt; reduce CapEx.
  • Automate to serve your mission
  • Encrypt (in motion and rest) Back-up  and Inherit (AWS best practice)
  • Enable your team with training (digital/classroom training, certification programme)
  • AWS Educate - fostering cloud talent. Free for children 14+
  • AWS Disaster Response (Connectivity at the edge, impact analysis, predictive analysis)

Jeremy Silver - CEO Digital Catapult

  • Network of organisations to accelerate early adoption of advanced digital technologies in the UK. Not for profit.
  • Working with start-up to governmental departments; raising £130m+ investment, industrial collaborations
  • Operation Bloodhound: UK mission to beat the landspeed record (Grafton LSR)

Now, on to the different tracks...

Management Track, Session 1

Presented by Jonathen Allen - Enterprise Strategust and Evangelist

The technology is in place to enable migration and it is becoming easier every year.

The Human element:

  • Moving to cloud affects careers
  • Fear and ambiguity for management and staff

This presents choices on how you proceed - hire new or train existing. Reference to Daniel Pink book "Drive". You then need to review the motiviation: Autonomy, mastery and purpose.

Allen then went through the 12 step programme, starting with Acceptance and setting out where the business is going - the leadership team needs to agree and explain. Remembering the change curve (some people are instant, others take longer).

  • Strategies (Create alignment, over communicate, spark motivation, developer capacity)

It appears AWS architects and evangelists subscribe to the 7 Habits of Highly Effective People by Steven Covey.

Explanations around the classroom training options:

Making a great point around providing employees with sandbox accounts (be it AWS or other services) so the team can play, try new things and most importantly: learn.

Moving on to the team composition "no team should be bigger than can be fed by two Chicago pizzas"

  • Product Manager / Lead Architect
  • Infrastructure engineers
  • Security engineers
  • Application engineers
  • Operations

Virtual team - dedicated to the task. Development, operations and delivery - DevOps.

Scale - split the team and bring in novices to work with the experts.

Start along the path to certification. Empower staff to train themselves and each other e.g. through A Cloud Guru (AWS Marketplace). Critical mass (10% certified) provides a psychological desire to adopt the same standards. Learning starts at the top (lead by example); recognise and reward expertise.

Management Track, Session 2

Laying the Groundwork for Change: Cloud Culture starts at the top by Norm Driskell, Director EMEA Solutions Architecture, AWS

Work on assessing the business drivers for change - time to market, profitability, business risk, employee engagement.

Analogies - bull fighter vs Jamie Chadwick (one has a team behind them and a plan themselves, the other doesn't)

Define - what is your cloud centre of excellence (CCoE)?

The CCoE is a multi-disciplinary team that is assembled to implement the governmence, best practises, training...

Your CCoE is the advisory and prescriptive function of the business - putting the "guard rails" on what your company works through and with what systems i.e. AWS. Focusing on

  • Knowledge (Industry Knowledge, Alliance programmes (supply / partner manager), Sales and tech enablement)
  • Experience (PoC / Pilot workloads, R&D Sandbox (give your teams the space to learn), Production deployments)
  • Offering (Pricing / Cross-charge (what's the internal cost and therefore external price), Go-To-Market strategy, path to procurement)
  • Operational Excellence (Governance, Automation - if you repeat it, don't do it manually, Cost optimisation - AWS reduces the price of products 20% or more.

Key takeaways: cloud practitioner is the way the learn. Make use of the AWS Well Architected Framework. Set organisational objectives to train and practise cloud implementation and management. Review security and compliance, as well as what services will be shared with which groups. It is imperative to continuously improve.

At Amazon, security is job zero

Automation of Disaster Recovery (DR) through AWS CloudFormation - you have new infrastructure available within minutes.

How do you get started with CCoE?

You need executive cover (even if you're a director of a company within a group). This person needs to have purchasing power (budget investment), time to learn, is active in the organisation and can recognise success or failure.

Think differently

  1. Change "the way we do things around here". Drive change from the top. Move from:
  • "Failure is not an option" to "Learning".
  • "Command and control" to "Decentralised" with guard rails
  • "Silos" to "Cross functional teams"
  • "Build and deploy in place" to "Automate"
  • "Long due-diligence" to "Adopt early and often"

2. Build the foundation - be the foundation.

3. Continuously iterate - you need to keep changing and adapting to the world around you.

Lunch time

A lot has been covered in a relatively short amount of time, so it's time for lunch. I'll give it to etcVenues, their food options are great. I wonder how much input Amazon had to this. Only complaint, not a lot of place to sit.

Management Session 3

Amazon's Culture of Innovation: Enabling Everyone to Innovate - Faye Holt

The mission - to be Earth's most customer-centric company.

The focus is on the customer, irrespective of culture, geography, technology. Starting with the customer and working backwards. What do your customers or clients need or want, and how do you get a product from this?

Holt shared a fantastic quote from Jeff Bezos's 2016 letter to shareholders:

Customers are always beautifully, wonderfully dissatisfied.

Amazon moved into multiple industries (books, CDs/DVDs, AWS, Kindle, Video, Groceries, Alexa/Echo, Bookstores and Go). Recongising innovation is hard, however Amazon is "stubborn on vision", but flexible on details. The how and when is less important, so long as you get to the end point.

Organisation for innovation:

  • Culture (customer obsession, hiring builders)
  • Mechanisms (encoded behaviours to facilitate innovative thinking)
  • Architecture (structure to support rapid growth and change)
  • Organisation (small, empowered teams)

Holt has shared there are "14 leadership principles" within Amazon. That's a lot to focus on; but she has picked on "Think Big" as one of the main principles.

The leadership principles are used in "every day language", so they are reiterated, reshared and engrained into the culture. They also want to share that they are willing to be misunderstood. The example given is the invention of the Kindle with the vision to provide a book within 60 seconds.

Decisions can be reversed - prefer calculate risk over long studies. This makes it a two-way door, rather than purely one-way. One-way doors, however, may include AWS Data Regions (this is a significant investment and not easily reversed). Two-way allows for mistakes and this is embraced - learn from mistakes and move forward.

Working backwards is key and Amazon asks five questions for each opportunitiy, beginning with "who is the customer". Writing a press release (one pager) with fictious dates and customer quotes to visualise how to get to the end. Creation of FAQs focused on two segments - customer and internal. Next up is the graphic visualisation of the end product - how will it impact the customer, what is the use case?

Amazon moved to microservices - faster and easier deployments. Enabling self-service platforms with gatekeepers. Using AWS, you have the option of using all of these microservices, some of them or a varied few. This has allowed Amazon to undertake 194 million deployments a year (around 6 per second per 30 days).

Back to the two-pizza team analogy: teams are fast and agile, fostering ownership and autonomy. Interestingly, AWS have differing views of what is the ideal number for a team; 5-10 or 6-12. The common thought, however, is that these teams must be decentralised and have the autonomy to build, run and then own their creations. The pizza teams allow:

  • Lower cost of innovation (small teams)
  • Reduce risk of repeat-failure
  • Encourage experimentation (if you know the outcome, it's not an experiment)

Example: Fire Phone

  • Failure (globally and public)
  • $170 million write-down of unsold stock

Lessons learned

  • The Alexa team came from the Fire Phone team (Echo, fireTV)

Management Session 4

Enabling Resilience through the Cloud: AWS Disaster Response  - Dominic Catalona

AWS Tenets

  • Improve
  • Invest
  • Train and support
  • Position

Programme is designed to help governments and not-for-profit organisations deal with natural or man-made disasters. Amazon has a Disaster Response Team, comprised of volunteers who are specialists, who can be deployed post-disaster with expertise in infrastructure, technology, data retention, communication. Each person must be certified in first aid (Red Cross trained). This service is designed to be philantrophic in nature; it's not to be profitable. The support provided is up to two weeks, but dependent on the requirements and partner.

Focus areas:

  • Technologies
  • Information
  • Disaster Response Action Team (DRAT...)

Featured use cases:

  • Rebuilding connectivity
  • Disaster Mapping
  • Building Applications for Good

AWS Disaster Response and AWS Snowball Edge

AWS Snowball Edge

Snowball Edge enables customers to run workloads on the edge, where connectivity is limited or non-existent.This also allows the DRT (Disaster Response Team) to work with the customer to make sure the technology works when and where it is needed most.

The team conducts field tests to make sure they have solutions to a variety of natural or man-made disasters. One test case was on St Kitts, enabling mission critical workloads to be run, especially during hurrican season where landslips and loss of connectivity are common.

Another example was working with the Hawaiian Volcano Observatory, who had decades of invaluable data stored in a local data centre at risk. Mitigated by use of an AWS Snowball, which transferred the data to AWS.

The DRT has worked with organisations through Help.NGO e.g. Global DIRT to provide Snowball Edge, S3 and EBS compatible storage to allow for processing and storing of drone footage (high-res, multi-layered images) to support Bahamas during Dorian. This has been documented by AWS.

This has resulted in Amazon creating and releasing the AWS Public Safety and Disaster Response Competency - how to best use AWS Snowball Edge, DRT etc. Part of this is linked to Project Resilience, where AWS offers up to $2,000 in credits to state, local governments, community and educational organisations in support of their business continuity plans in AWS.

Afternoon Break...more food

There's so much food here - I can't believe anyone could go hungry at these events.

Tech Track Session 5

Compute at the Edge - Martin Bishop and Wayne Soutter

As you can tell...I decided to switch it up and go to the tech track for the last separated workshop of the day.

AWS focuses on a low-latency, high fidelity cloud experience. This is especially important in content production / distribution (gaming), TelCo operations, autonomous vehicals, financial services. This leads into the next topic - AWS Outposts.

AWS Outposts

This would be like moving an AWS Region into a customer's data centre (or co-located DC). Doesn't sound very "Cloud like", but is as close to it as possible; you're still using HaaS (hosting as a service) and AWS as you would if it weren't co-located. The Outposts are optimised for maximum service for minimal footprint. They support ECS, EKS, App Mesh, EMR, RDS and soon, S3.

Outpost is monitored; AWS and customer has full access to health metrics and alerts. Any hardware faults can be easily resolved as the Outpost is modular. It offers local and parent region access. It can be installed and shipped across NA, EMEA and APAC; exception is not Mainland China, but Hong Kong is available.

Pricing examples are dependent on capactiy and type, with options to make full upfront, partial upfront or a monthly fee. An example of all upfront payment is between ~$533,000 to $1,000,000+.

Local Zones

This is an extension of an AWS region, to get even closer to end-consumers. The first release will be in Los Angeles. This is still managed by AWS.

Connectivity over mobile networks - AWS Wavelength

The current approach can result in latency >100ms. AWS aims to reduce this as much as possible with AWS Wavelength, by embedding compute and storage inside a TelCo's 5G Network. This enables developers to provide a service in <10ms latency.

IoT on AWS

Setting up, running, securing and managing Internet of Things (IoT) devices. AWS has a way to manage and analyse the IoT stack, focusing on revenue growth and operational efficiency. Customer example is Centrica / British Gas, using the Hive IoT platform (temperature and other services) - new business / revenue model. Deutsche Bahn is using AWS IoT for operational efficiency (e.g. predictive maintenance). A wide range of applications which can be managed through AWS, with multiple examples (some shown in the gallery below).

A great example was the implementation of LPWAN (Low-Power, Wide Area Network) in place by Newport, Wales, which is used in part to connect IoT devices related to flood warnings.

Why is this happening now? The cost of sensors, compute time and connectivity is decreasing over time. There are now around 28 billion IoT devices active in the world. This is why AWS is offering a service to manage all aspects of IoT - SDK, device management, security and analytics.

  • AWS Greengrass is used for setting policies (what, when, prioritisation) of IoT services.
  • AWS IoT Core is used for creating multiple endpoints within a single AWS account, but have a unique configuration for each one. Custom domains and authoriser. Alexa Voice Service is also integrated to provide built in capabilities to reduce the compute and memory footprint required by ~50%.
  • AWS IoT Device Management provides secure tunneling for IoT devices, irrespective of firewalls and corporate security policies.

Final Session - Live demos and quiz

Lead by Sunil Mallya (Principle deep learning scientist, AWS), Tim Hinchley (Head of Cloud, DVSA) and Craig McCallum (Senior Solution Architect, AWS).

DeepComposer Demo

Mission - to put machine learning (ML) into the hands of every developer.

Prediction by AWS - 58 million jobs created by ML and AI in the coming years. To assist with this, AWS is running educational training, certification and also devices. There's also explanations of the different type of ML - supervised learning, unsupervised and reinforcement. AWS works to generate samples and data sets used in ML, using Generative Adversarial Network (GAN). This also provides the option to feedback to the ML via the descriminator. The reason for the model is to be able to tell how effective the ML has been at creating something e.g. is this jazz music actually jazz music?

There are plenty of examples of practical uses of Generative AI - Autodesk (Airbus, NASA), scientific/medical labs e.g. Gildewell Laboratories who manufacture dentures. Video game assets are also able to be generated using GANs.

You may be able to tell - ML and AI is not part of my knowledge base. Yet.

Thoughts and wrap-up

In short, I wish I had been able to

  1. Go to AWS re:invent originally
  2. Go to both sets of sessions (tech and management).

I don't regret choosing to go to the majority of the management tracks, only missing out on the cost optimisation (which would be useful..), but I do wish this had been across two days. I'll definitely be downloading all of the slides and am already signed up more, future sessions.

Key learnings:

  • AWS Staff are great at selling their belief in the technology and company as a whole
  • There's a lot of great technology baked into AWS and it's a continual, iterative approach to architecture
  • Culture is big within Amazon
  • Didn't know about the DRT section and Snowball, very cool, philanthropic venture
  • AWS is not afraid to make or admit to making mistakes, so long as they can show they've learned from it

Today has certainly renewed my desire to finally finish my A Cloud Guru training videos to take my Associate Solution Architect exam on AWS, but has also made me want to review Google Cloud, their similar sessions and certifications. At the end of the day, I want to find a way to implement both with my work and make them reliable, revenue streams. To finish, a picture of me at the conference - I haven't been to many so far as my previous employer focused on sales / marketing attending these, but I'm happy to report, this is going to be more common in my new role. Thoroughly enjoyed today.

Author image
All views are my own. Proud fiancé and cat dad, Technical Director at We Are Kitty