Scaling innovation @ AWS re:Invent 2016

Luca Bianchi
10 min readDec 3, 2016


Las Vegas becomes a kind of interesting place during Falls even if you are not interested in shows, girls or gambling: beside casinos and lights, you can find the largest Amazon Web Services conference of the year, the so-called AWS re:Invent.

Amazon re:Invent started on Monday, 28th and will bring technology and innovation to Vegas until next Friday. The first effective conference has been Tuesday, 29th and on stage, many APN partners presented their solution and special use cases of cloud and edge technologies

The first day started with a compelling Workshop about Alexa Skills, a core theme for AWS and probably one of the most relevant due to the upcoming announcements. If images mean anything, this should clear every doubt about the importance AWS is giving to Alexa and voice/conversational interfaces.

That was an entry level workshop about Amazon Alexa, the voice engine provided into Amazon Echo (and they provided one Echo Dot to each re:Invent participant) that is fed through JSON Skillset model providing utterances and intents to Alexa service.

The most important thing is that Alexa service does not require echo device: it can be integrated everywhere, also on a RaspberryPI. To start building a skillset, checkout

There are also a couple of projects that can help you waking up Alexa from an hook word: and

You can also try online simulation with EchoSim:

Moving on edge technologies, PwC presented the use of Blockchain algorithmsinstead of traditional databases to manage transactions and decentralized evaluation of operations. This technology (which is the ground of Bitcoin platform) provides a database which is immutable by design and transactions are stored and dragged along, providing the capability to rebuild the complete history of every transaction. This opens to a whole set of use cases in sectors such as bank and insurance companies to validate transaction in a secure way. Two interesting projects are Ethereum and Monaxproviding support for Blockchain nodes.

Then the day moved to microservices topic with Session Advanced Techniques for Managing Sensitive Data in the Cloud and Session Building Complex Serverless Applications that outlined some best practices adopted by APN partners such as Nike when dealing with Microservices, especially if in a serverless environment.

Tuesday night ended with James Hamilton providing deep details about AWS infrastructure and the new regions that are going to extend support in Australia, Canada, and Africa next year.

The second day started early with Andy Jassy (AWS CEO) on stage, announcing some great news in cloud computing and more than ten new services. It all began with the superpowers using AWS will provide to your business, in particular:

  • Supersonic Speed: With AWS, you can move from idea to implementation faster than ever before.
  • X-ray vision: With such power, you can understand your customers and your business better through Analytics, and eventually see the meaning inside your Data.
  • Immortality: If you want your business to be around for the long haul, you will not be able to do it without evolving technologies.
  • Flight: The freedom to fly, to build fast, and to better understand your data.
  • Shapeshifting: It is no longer just a choice between on-premises and the Cloud; you can also go Hybrid.

On the computing side, now AWS has six new instance types ranging to standard T2 general purpose, to new highly specialized FPGA instance.

On the service side, AWS announced Amazon Lightsail, to fire up and handle Virtual Private Servers very fast and Amazon Athena, an interactive service that makes easy to query S3 for data within seconds and elastic GPU extension for every EC2 instance.

The main topic of the day, however, has been the announcement of a brand new category of services: Amazon AI. In particular, we have been able to try:

  • Amazon Rekognition: a manage image recognition, face detection and face search service. Only REST, CLI and Java SDK supported right now, but we expect every updated SDK in the next few days.
  • Amazon Polly: is capable of supporting text-to-speech implementing Deep learning, starting from a textual string generates an MP3 localized to the user language and way of saying.
  • Amazon Lex: is the conjunction of two different fundamental components of the Alexa Engine: the Natural Language Understanding and Automatic Speech Recognition that can provide Voice interfaces and conversational (step-by-step) navigation to your apps. Currently in Limited Preview.

When you think all the cards are on the table, Andy Jassy drops one of the most awaited and interesting news of the last two years: AWS Greengrass! AWS Greengrass will let you embed Lambda Compute in connected devices, plus secure messaging and data caching as well. You will be able to define Lambda Functions directly in the AWS Console with the same programming model and deploy them into your IoT devices.

Last but not least, Andy announced updates to Amazon Snowball supporting edge locations and the new Amazon Snowmobile: a truck full of Snowballs, to handle massive cloud migration.

After such big news, you could think that the day would proceed smoothly towards the evening Pub Crawl, but that was not the case: a Session on architecting next generation SaaS Applications provided a clear understanding of the topics to address when building multi-tenant platforms on AWS in terms of user management, tenant identities, isolation, and segregation.

At the end of such great day, the team moved to the traditional Pub Crawl and spent a few hours finalizing our hackaton project on Alexa developed in just two days with our friends from beSharp.

If you want to learn more about our Alexa hackaton project, you can: we’ve released the complete source code on GitHub.

Day Three came with Dr. Werner Vogels opening keynote and did it! After a “cold start” with a brief introduction on the landscape of Microservices and Twelve-Factor App principles, he went straight to the point recalling the vision AWS has Lambda functions in the ecosystem of cloud environments.

Werner talked about transformations and the way IT is changing at an unprecedented pace thanks to cloud computing, in particular, they occur in Data, Development, and Compute, putting developers at the core of this new age of IT.

On the Development side, the focus is on agility, a paradigm that should drive releases cycles and teams. AWS plans to support this with a set of new services such as

  • A fully managed Chef server named AWS OpsWorks for Chef Automate
  • A collection of EC2 management tools for task automation named EC2 Systems Manager
  • AWS CodeBuild: a fully managed build server for compiling code and running tests. This service is one of the breaking news of today because together with AWS CodeCommit and AWS CodePipeline provides finally a fully managed CI/CD environment for startups and enterprises. Codebuild supports many languages such as Ruby, Python, Java, C# and NodeJS
  • AWS Personal Health Dashboard is useful to build personalized dashboards of AWS services health monitor that will benefit customers of your SaaS product. Joining this product with the release of AWS Serverless Developer Portal on GitHub and the upcoming API Gateway Marketplace (both announced by Tim Wagner in his session) represent a huge step forward for companies providing multi-tenant API based solutions on AWS. A must-have for every software vendor of the 21st century.
  • AWS Shield is a partner-killer because it offers out-of-the-box DDoS protection for apps and services. A great new for developers and companies (definitely making the life difficult for many AWS partners such as Trend Micro, Acquia, Sophos, Barracuda, and others.). Matching AWS Shield with AWS WAF provides a complete managed solution for cloud security.
  • AWS X-Rayis another partner-killer service on the cloud monitoring domain: it aims to analyze and debug production, distributed applications, such as those built using a microservices architecture using a transparent approach. The service is in limited preview, thus Lambda is currently supported only partially, meaning that X-Ray monitor must be explicitly included, but it is astonishing in the number of details it is able to extract from our infrastructure, especially when dealing with microservices.

On the Data side, Vogels presented three brand new services tha can change the attitude of companies towards data analysis and user engagement:

  • Amazon Pinpoint is a fully managed campaign service that can open new scenarios to developers and SMB companies at a competitive cost. We’ll talk more about this later.
  • AWS Glue is a data catalog service with ETL (Extract, Load, Transform) capabilities, natively integrated with S3, RDS, Redis and JDBC databases. Since Vogels pointed out that 80% of work done on data analysis is not analysis at all, AWS Glue aims to become the missing tool to manage distributed data.
  • Amazon Batch is a simple, fully managed and really needed batch manager engine that can handle priority based queue and any kind of task on AWS resources or customer data.

On the Compute side, AWS announced

  • AWS Lambda support for C#
  • AWS Lambda@Edge: You can now run AWS Lambda functions at CloudFront locations, which allows you to process data at the edge locations to achieve low-latency request/response.
  • AWS Step Functions: is an automation addendum to Lambda world, providing components coordination visual workflows. It is already in General Availability, this meaning the road is open to Lambda functions orchestration, achieving complex behavior such as concatenation, parallel execution, retry policies, long-running tasks, and so on.

And this was just the morning! Following the day we’ve able to dive deep into these innovative services

Session: What’s new in Amazon Lambda presented a whole set of upcoming changes to AWS Cloud nervous system. One of the most significant outcomes of this sessions has been the idea that Lambda is to Cloud what bash scripting is to traditional servers. Stated this assumption, comes clear that Serverless Computing is just a small part of what Amazon thinks about Lambda. Tim Wagner talked about the recent release of SAM support in CloudFormation that together with CodeBuild and CodePipeline can provide a full CI/CD solution for serverless projects, with the fantastic addition of managing NPM packaging on the deployment side, a long-awaited feature of Lambda that now can be achieved within the ecosystem. Moreover, Lambda supports the newly announced X-Ray technology as well as the just released C# development language. During Q&A Tim also uncovered that Node 6 and Go support are coming in the next months, however, no news about other languages such as Swift, being postponed depending on their diffusion. Finally, Dead Letter Queue support for Lambda events improves resilience while the upcoming release of SQS triggers (in the next months) can provide minimal support for message bus communication between Lambda services.

Session: How to enable Real-Time Mobile App Engagement with Amazon Pinpoint is a brand new service that can support customer engagement, lowering the estimated cost of 10–15$ per user, a very high amount considering the high level of uncertainty about users conversion to customers. To reduce such massive investment, a lot of companies (Adobe, IBM, and many others) spotted push notifications as the perfect channel to handle customer engagement throughout all the app lifecycle. Amazon Pinpoint is born to be a cost effective campaign management solution for developers, startups, and enterprises of any scale. Pinpoint supports plain notifications through Apple or Google (web push and other channels coming in the future) providing an out-of-the-box management layer. It is incredibly easy to define, through APIs or Mobile Hub console, labels representing user segments, which are updated real-time from mobile app interaction. Whenever a marketer decides to start a campaign, she simply has to choose a cluster, define a particular message and schedule id accordingly to user local timezone and customized with her specific parameters. Another interesting capability of Pinpoint is the support for A/B testing of a campaign message and the capability to deliver rich media push notifications as well as silent push notifications directly to Pinpoint native app SDK. Amazon Pinpoint can be a significant leap forward customer engagement, easily integrated with custom logic defining customized content for campaigns.

Session: Building Distributed Applications with AWS Step Functions makes easy to coordinate functions and microservices applying the metaphor of state machine backed either by GUI console and JSON coordination format. AWS did more than this because defined an open standard for function state machine management ( Moreover, Step-functions is not bounded to Serverless Computing but supports everything that performs computing on AWS by definition of tasks withing the state machine and their execution. You can think of SF as a workflow automation tool to handle and manage your AWS cloud.

AWS re:Invent is almost over, but its outcome is going to shape the landscape of Cloud Computing within the next few months. It has been a two-day full of announcements and innovation here in Vegas and, this time, what happens in Vegas, won’t stay in Vegas!



Luca Bianchi

Chief Technology Officer at Neosperience. Loves speaking about Serverless, ML, and Blockchain. ServerlessDays Milano co-organizer. Opinions are my own.