re:Invent 2019: A Very Late Recap on Tech Highlights, Conversations, Observations & News
Like last year, re:Invent in Las Vegas in 2019 this year was nothing short of a fantastical, epic event. While attending, one is left thinking no other technology matters beyond what is represented at the conference. The hype, the excitement is tangible. My mind has finally settled since leaving the conference and I’ve been able to remove the hype from my observations and I wanted to share those thoughts here (even though they’re ~3 weeks late.)
Quick observations from conversations
Hours and hours of each day at the conference were spent in working sessions or conversing with industry enthusiasts, some even there to present in the Expo at vendor booths. Some of the conversations were crap, but some definitely expanded my view of an already expansive industry.
Observation #1: Last year enterprises went all in on serverless, many hard lessons learned, and AWS countered with improvements to their services. It was exciting to see so many companies embracing serverless and so willing to share their failures or lessons learned. It’s clear that AWS is listening to the industry front-runners and is also going to continue investing heavily in serverless technologies.
Observation #2: There were very few and rare wins with true enterprise “lift and shifts” to the cloud. Mostly what I saw were failures when companies were trying to do pure “lift and shift” migrations. The true and visible wins were when companies “strangled” out components with serverless re-thinking on their monolithic applications or platforms.
Observation #3: Machine Learning is looking to be a key differentiator for most industries represented at AWS, not just healthcare. I’m currently in healthcare, and many companies are finding some real wins with ML but have only just scratched the surface. Companies are going to need to invest in ML to stay competitive in their industries.
Observation #4: Docker as a brand is dead. This kind of hurt my heart, being a huge fan of Docker. Maybe dead is too harsh, but the mention of Docker specifically was rare. Containers were everywhere and containerization technology, but it was always focused on Kubernetes (aka k8s), not Docker. K8s is king and has won the container orchestration wars. Also, the trend and theme for container orchestration needs to be simpler.
Observation #5: Vendors want people and teams to think that everyone is doing “beyond DevOps”, meaning Git + GitFlow + advanced DevOps. If one isn’t careful, they’ll come away thinking their own work teams are far behind everyone else and feel disheartened. Cutting through to the main point is that teams are pushing aggressively towards automation, logging, monitoring, and ensure data insights are on steroids. Focus on the DevOps basics, ignore specific tooling, and teams will be fine.
Observation #6: InfoSec activities and tools were prevalent and more visible than I’ve seen in the past. I love information security, and it’s awesome seeing that efforts are being shifted left in the development pipeline, meaning that security is more and more being considered as value delivered during the development life-cycle, not sometime after it’s completed. Having baked-in security embedded in teams “definition of done” is going to have great impact on the technology industry.
Observation #7: Migration tools to help companies get from on-premise to cloud are getting very advanced. It’s clear that large data-driven enterprises are taking the public cloud serious. There are some very advanced tools now to help those interested companies make the big leap (or lessen the chasm size of the leap) in getting their data and applications to the cloud. I saw many Oracle database schema and migration tool demos and they didn’t disappoint.
Observation #8: Tremendous emphasis on re-thinking how we do architecture in a serverless world was the topic for many working sessions that I attended. Monolithic mindsets will result in monolithic architectures in the cloud without any of the benefits. TL;DR migrating to the cloud won’t work if we don’t architect for cloud.
Observation #9: Education & continuous improvement is key to keeping up in our industry. I tell this bluntly to my teams at my employer: one cannot hope to stay technologically relevant unless extracurricular training and education is embraced. So much content, so many pivots, so many improvements will render a technologists knowledge obsolete unless one treats that knowledge like a sword going to battle: it’s going to need to be constantly sharpened and tuned.
Like everything I author, these are only my observations. They are not the thoughts of my employer or my colleagues. They formed from walking hours and hours through re:Invent (~25K steps per day for 5 days.)
On-premise hybrid solutions
Some of the most exciting pivots I’ll be watching for in 2020 are around the many new hybrid offerings coming out from AWS, which arguably gives them largest hybrid footprint in cloud wars.
While AWS doesn’t want to use the term “hybrid” a lot, I think enterprises understand that it means they can extend their AWS experience to on-premise or close to on-premise compute and storage.
AWS announced three huge capabilities here that are important, including going GA on Outposts and announcing Local Zones and Wavelength. Why does this matter?: AWS took the hybrid idea and doubled down on it and it’s going to pay off.
For customers who want a low latency experience on-premise with Outposts, lowest-latency in the public cloud with Local Zones, or in the core carrier network with Wavelength, AWS has us covered.
Machine learning & artificial intelligence
The nerd in me was stoked about the announcements on Graviton2 for EC2 M, R, and C 6th Generation instances. Based on an Arm N1 core, AWS says these new instances deliver up to 40% improved price/performance over comparable x86-based Skylake instances. This is technology specifically created for ML and AI.
Inf1 EC2 instances with Inferentia were also announced. Last year, AWS pre-announced Inferentia, its custom silicon for machine learning inference. This year, it announced the availability of instances based on that chip, called EC2 Inf1. This is more technology specifically created for ML and AI.
“No ML experience required” AI services were in abundance in working sessions, keynotes, and vendor booths. AWS came out strong touting new services that don’t require ML experience. Kendra, CodeGuru, Fraud Detector, Transcribe Medical, Augmented AI. Why does this matter? Lowering the lift of learning and embracing ML in enterprises makes it much easier to implement.
SageMaker Studio was also released. Amazon SageMaker Studio is the first comprehensive IDE for ML, allowing developers to build, train, explain, inspect, monitor, debug, and run their machine learning models from a single interface. Having a unified developer experience will drastically reduce the ramp up time for developers to learn ML, AI and the complex toolchains that currently exist.
Containers, serverless & databases
Honestly, I was really disappointed in the container-centric announcements at re:Invent. The big one was that AWS Fargate for EKS is available now (serverless EKS/k8s.) Not to be outdone by standard EKS or ECS, serverless fans now get a way to more easily deploy, manage, and scale k8S on AWS. This is awesome, but it was really the only container announcement that mattered, beyond the FireCracker and Nitro stuff.
Unlike in the container realm, there were many, many serverless enhancements. AWS API Gateway v2 for HTTP (reduced cost API Gateway), RDS Proxy, Provisioned Concurrency for Lambda functions, S3 Access Points, Eventbridge Schema Registry for event-driven architectures, and more were all let loose on the world. This stuff was exciting and helped me to forget about the lack of container news. Serverless is here to stay, evolve.
AQUA for RedShift was one of the more interesting database announcements. AWS is using a hardware accelerated cache bringing ASICs and FPGAs to RedShift resulting in what it says is 10x better query performance than any other cloud vendor. It’s awesome to see where RedShift is pushing the data warehouse paradigm. It’s even more exciting to see how this is pushing competitors to compete.
AWS unveiled many database improvements, like Cassandra. AWS enters the Cassandra space with this announcement. While the use cases for Cassandra are like those of the Amazon DynamoDB platform (both are distributed databases), the choice will likely be driven by developer preference.
Bringing ML to Aurora: AWS also made announcements extending its Aurora transaction and RedShift analytic databases. The RDS technologies behind Aurora are more easily integrated with ML and AI services provided by AWS. This is more evidence that ML and AI are going to be integral differentiators in all technology augmented industries.
Many RDS & Aurora MySQL/PostgreSQL enhancements & upgrades were announced that I couldn’t keep up with. See the full list of announcements for the complete listing.
Education, training & sessions
As mentioned in Observation #9, one cannot hope to stay technologically relevant unless extracurricular training and education is embraced. For those interested, here are collections of cultivated sessions centered on the technologies in each header.
Overall AWS re:Invent recorded sessions / guides/ training
Getting started with AWS, period
Security guides / training
Container & serverless guides / training
Machine learning guides / training
Database learning guides / training
DevOps learning guides / training
Startup learning guides / training
To summarize, my mind was blown again just like last year at the amount of content delivered. I take every opportunity to be a #PerpetualLearner and even still I find myself struggling to keep up. I love serverless and ML. I also love how fast our industry is evolving. My hope is that some of these observations help someone else improve on their pursuit of being a #PerpetualLearner.