Top 5 Takeaways from AWS re:Invent 2019

I attended my first AWS re:Invent this year. It attracted 65,000+ people from enterprises large and small in a variety of industries and countries. I had a front-row seat at a fast-growing public cloud company while I was at Google Cloud. But the scale that AWS operates at is different and impressive. They announced a slew of new services (175 total now vs 140 last year) with a prominent line up of CEOs such as Cerner’s and Goldman’s talking about how AWS helped with their digital transformation.

We, at AV8 Ventures, strongly believe that the tectonic shift to the cloud is in its early innings, and it will lead to many exciting startups in the next years. 

Here are my top 5 takeaways from AWS re:Invent 2019:

1)    Security is one of the core tenants of public cloud:

Considering the fragmented nature of security landscape and complex nature of new infrastructure architectures, there’s a growing interest in public cloud security capabilities. Moreover, security has been a key barrier for large enterprises to fully embrace the public cloud.  There will be companies adopting best of breed solutions. But a big chunk of enterprises will rely on AWS and it’s tightly integrated partners for security solutions. AWS is doubling down on this trend and announced a number of interesting products to address security of sensitive data, applications, and misconfigurations, which create large vulnerabilities i.e CapitalOne breach.

In this changing environment, how do you manage privileges and compliance across 100s of AWS accounts, 1000s on instances, billions of S3 objects, trillions of DynamoDB? entries? How do you avoid overly permissive deployments and give the right access to the right things at the right time? You need powerful guardrails for developers to stay secure. The following key product announcements will have an impact here:

a)     AWS Identity and Access Management (IAM) Access Analyzer ensures that only the intended access to resources is allowed through security policies.

b)    Amazon Detective helps with root cause analysis and investigation of security incidents.

c)     While it’s not new, Amazon Security Hub, which is a single place that aggregates and prioritizes security alerts from multiple AWS and partner services, is now critical to integrate with.

2)    AWS wants to own the end-to-end ML development workflow:

AWS addressed growing ML development interest in the enterprise world with a full suite of ML capabilities such as Autopilot to train & tune models, Model Monitor for ML observability, Notebook to manage workflow notebooks, and others. One that caught my attention was Sagemaker Studio, which is an IDE for ML that allows data scientists to streamline model developing, training, deployment and testing. I’m curious to watch the adoption of this service as data scientists are bombarded with many new solutions from exciting startups, and Sagemaker Studio needs to convince them the benefits outweigh the learning curve. One key advantage AWS has over GCP in ML is the data gravity. Data scientists and developers will find ways to work where data resides. Andy Jassy mentioned in his keynote that “85% of TensorFlow in the cloud runs on AWS” despite the fact that Google developed Tensorflow.

In addition to ML platform and infrastructure services, AWS is not shy about introducing pre-packaged ML applications competing with its partners who have a love and hate relationship with AWS.

• Amazon Kendra – ML-powered enterprise search.
• Amazon CodeGuru – ML-powered automated source code review and software performance suggestions on latency and processor utilization. They claim existing products were unusable with 70+% false rates.
• Amazon Fraud Detector – ML-powered identity and payment fraud detection fully-managed service.
• Contact Lens for Amazon Connect – ML-powered contact center analytics on customer interactions, sentiment analysis, and trends.

3)    AWS is focused on edge & hybrid vs multi-cloud capabilities:

In addition to cloud applications, AWS also wants to address on-premises applications that may need to live close to local data sources or need to run with low latency. Different than other vendors, AWS didn’t talk about multi-cloud which could help with certain aspects such as avoiding vendor lock-in, risk mitigation, pricing power, and others. But, AWS’s focus was the increasing convergence between on-premise and AWS i.e. hybrid infrastructure. There were three key announcements here, covering all the bases for enterprises:

• General availability of AWS Outpost which is fully-managed AWS-designed racks to bring some AWS services (RDS, EC2, EBS, EKS, EMR, and S3) to the edge with AWS or VMware control plane. AWS will deliver, install, manage & support these racks ?? They will do site surveys, find the right form of power, cooling, etc. This is not a trivial undertaking but strategically a good move to lock those on-premise applications to the AWS ecosystem. Now all three cloud providers have some form of a hybrid solution: Google Cloud Anthos and Azure Stack.
• AWS Local Zones is a new way to run low-latency required applications closed to local workloads. First one is available in LA, driven by VFX studios and media companies. Google Cloud announced something similar last year too.
• AWS Wavelength is way more interesting than Local Zones I think. They enable developers to build super low latency applications (ML at the edge, industrial IoT, etc.) by bringing AWS compute infrastructure to the edge of the 5G network. There are few exciting startups such as Packet and Vapor IO who have been building similar edge computing nodes but in the wireless towers.

4)    AWS Partner ecosystem is massive and impactful:

While many partners have a love and hate relationship with AWS, they continue to run their applications on AWS and partner with AWS on go-to-market. The Expo Hall was full of impressive enterprise technology companies and startups, showcasing/announcing their products on AWS such as Instana’s support for Node.js on Lambda or Dynatrace’s support for AWS hybrid deployments.

As someone who ran strategic partnerships at Google Cloud, I was especially impressed by announcements around two of these partners:

•  The AWS/VMware partnership is critical in the hybrid cloud world. VMware is well-positioned in the world of multi / hybrid cloud architectures driven by their strong product portfolio and smart set of acquisitions such as Heptio, Pivotal, Nicira, Carbon Black and others. They already have a strong presence in VM-based workloads in the on-premise segment. They now utilize Kubernetes to help enterprises with containerized applications in their migration to the cloud. AWS worked closely with them and push strong joint solutions such as VMware Cloud on AWS, RDS on vSphere, and is now pushing VMware Cloud to AWS Outpost. This gives AWS a strong advantage in migration opportunities.
•  While it’s early, AWS’s non-exclusive partnership with Verizon for AWS Wavelength could be equally important considering the potential of 5G. Verizon CEO Hans Vestberg joined Andy Jassy on stage to announce this service and had him talk about the exciting upside with low-latency edge applications.

5)    Oracle database migration gave AWS huge credibility:

I really liked the session on Amazon’s journey on getting Oracle out of the building. The session started with: “Managing 75PB of data growing at an exponential rate is expensive time-consuming. Some teams saw 90% reduction in operations costs with AWS-managed database services.”

Which CIO wouldn’t pay attention to this case study?

In addition to these top 5 areas, there were many exciting announcements too.  Graviton2, Arm-based processor, and Inferentia ASIC, ML inference accelerator are great additions to the large family of compute instances and will help with price/performance. I also liked ML-powered AWS Compute Optimizer which helps developers optimize compute resources for their workloads and more importantly their AWS bill ??  Then, there’s AWS Braket which is a fully-managed service for data scientist to use quantum computing. I guess they needed an announcement to match Google’s quantum supremacy few weeks ago ??

Overall, I’m impressed with what AWS has done with re:Invent. ?????? They will continue to be a juggernaut in all things cloud. Our team at AV8 focuses on many areas of innovation in the enterprise stack all the way from infrastructure to application layer. If you are working on an ambitious enterprise startup, please reach out to us!

PS: Headline photo and some product descriptions are from AWS content.