As the pace of innovation continues to accelerate, staying ahead of this wave is something that both challenges and inspires us as software consultants. After a great week with the SpinDance Engineering team in Las Vegas for AWS re:Invent 2018, we’ve come back inspired and excited to share our experience and what we learned about the new tools and technologies with our co-workers and customers. Staying up to date with all of the AWS offerings can feel like a full-time job, and attending re:Invent is a great opportunity to learn in abundance from other engineers and AWS partners. It’s also a great opportunity to talk with and sit down with the engineers that help build some of these amazing tools.
Below are some of the trending topics and themes we brought back.
IoT Continues to Drive Ahead
With announcements like IoT Events, IoT SiteWise, IoT Things Graph, IoT Device Tester, ML inference at the edge with Greengrass, the exploding curve of new technologies continues to drive forward for IoT.
• IoT Events supports the creation of virtual state machines to allow the detection and handling of telemetry events across any number of devices.
• IoT SiteWise supports the collection and organization of site-wide device data. This service was born out of many common problems in the industrial and manufacturing space and is another example of AWS packaging up a solution to a difficult problem observed in real-world applications.
• IoT Things Graph allows the orchestration of various workflows and automation across many devices on Greengrass.
• IoT Greengrass ML Inference continues to enforce the theme that processing power continues to move closer to the edge. This power is a theme well beyond AWS and re:Invent, but it was reaffirmed at the conference.
• IoT Greengrass Security received a boost with a few new features, including the ability to integrate with Hardware Secure Elements and to leverage rotating keys provided by the AWS Secrets Manager.
AI and Machine Learning
Machine Learning (ML) was another huge theme of the conference. Most sessions mentioned how ML could be integrated into a particular architecture solution, in some cases to the point of absurdity. Many sessions displayed the three-tier AI stack that AWS offers with the high-level AI services and abstractions like Comprehend, Rekognition, Translate, Lex, Transcribe, and Polly at the top level. The managed ML service, SageMaker itself makes up the next level tier. The supported frameworks like TensorFlow, PyTorch, and Apache MXNet make up the bottom level along with the supporting infrastructure including the new P3 and C5 EC2 Instances. From IoT to Stream Processing, to Augmented Reality/Virtual Reality (AR/VR), Machine Learning was mentioned in every context. Overall, with the expanded support of the AI tooling, ML has been made much more accessible to the AWS developer community.
More Purpose-Built Databases
Modern applications have created new requirements. We expect millisecond level responsiveness, support for millions of users across the globe while maintaining predictable performance. These requirements have driven changes in how we build applications and the data technologies we use. AWS continues to build on the variety of data storage services, each intended to do the heavy lifting for particular use cases. Big changes were announced for these ‘purpose-built’ databases. Following up with the success of Neptune, the graph database that was announced last year, both Quantum Ledger Database (QLDB) and Timestream expand the suite of purpose-built databases:
• Quantum Ledger Database uses an immutable, transactional log that makes up complete and cryptographically verifiable histories of application changes in a system. QLDB is highly scalable and is another serverless database offering from AWS.
• Timestream was also announced, adding a true time-series database to the suite of database offerings. Timestream is a great choice for storing and querying operational, sensor and other IoT data due to the efficient storing and processing of entries based on time.
From IDE tooling support, the new Ruby Lambda Runtime, and now custom runtimes (including reference implementations of Rust and C++), the way we develop serverless business logic is evolving for the better. There is nothing worse than managing a sea of lambda functions without an elegant way to reuse pieces and work with them in a comprehensive repository. The changes introduced with Lambda Layers, and the IDE tooling support show they are investing in moving this sort of development forward to be as mature as other development methodologies.
New IDE Tooling Support:
• Visual Studio Code
• C++ (custom)
• Rust (custom)
Logistics, Logistics, Logistics
Having this conference on the Las Vegas Strip with over 50,000 attendees presents a logistical challenge, and potentially a nightmare if not orchestrated with precision. As a first time attendee, I was impressed with how the event organizers set all participants up for success. Las Vegas by its nature is an environment that is luring and distracting … casinos, shows, shops, etc. The signage in the resorts can be confusing and often takes you the long way around, past many of the attractions on the way to your intended destination.
Imagine dropping the entire population of a major university onto an unknown campus and hoping to feed, herd, and provide an overall outstanding experience to everyone. It is quite a challenge, and the AWS re:Invent organizers build all of the necessary pieces needed to make this work well. Here is how they did it so well.
Support staff was distributed throughout the campus wearing flags to stand out from the glitter and blinking lights. Pedestrian traffic control patterns were developed, keeping things in the most common traversal points flowing smoothly. Point-to-point shuttles were available to cart us from location-to-location on what was typically full days of sessions and workshops, and the mobile conference app had real-time travel times between venues. Imagine what it takes to herd thousands of people showing up in the basement of the Venetian (shown above) for breakfast and having lots of small warm buffets distributed and ready to go with low wait times. The logistics of the event were very impressive. It’s really hard to capture just how big this room is … we referred to this as the “hall of infinite breakfast.”
Conclusions and Take-Aways
Attending events like re:Invent can make a huge impact on engineers and company teams. We love to learn. Coming together with so many other customers and engineers to learn together generates great excitement and collaboration and drives and inspires us to high standards in our daily work. As consultants, it’s our job to know the right tools for the right job to ultimately drive the right value for our customers. It’s important to take the time to break away and focus on learning about the tools and technologies that drive our decisions. I came back with three major takeaways:
1. Solutions for common IoT use cases are being packaged up into new AWS offerings to jump start customers and developers in this space.
2. Machine Learning is becoming a discussion point for many different contexts, some valuable, some not. All in all, it’s becoming more accessible to developers on AWS due to the widespread integration possibilities.
3. Serverless will continue to advance and change the way we architect and develop. The approach to Lambda development is evolving, and the tools are starting to make the development ergonomics far better.
We were lucky to have four representatives from SpinDance this year, and while we spread out at the conference to get coverage on a wide variety of topics and sessions, I know we’ve all come back with the same fire to Werner Vogles always says “Go Build.”
If you are ready to build, ready to plan, or just ready to talk about what’s possible, reach out to us at email@example.com … this is what we do best.
re:Invent Conference Resources
Here is a listing of some of my favorite sessions and topics, including slides and the session videos when available: