AWS Continues to re:Invent Cloud Computing
Andy Jassy took center stage Tuesday for Amazon Web Services (AWS) annual and first ever virtual re:Invent conference. Andy’s keynote was jam packed with new product announcements and feature enhancements with compelling customer stories from all industries sprinkled in between. There were so many major announcements that I had a hard time keeping up with it all and I’ll do my best to summarize some of the highlights below and provide you with links to drill down for more details.
Jassy opened the keynote talking about the State of the Cloud
Referencing a Gartner report, Jassy declared that AWS is running 45% of all public cloud workloads which is 10% more than the next 4 cloud providers combined.
He then highlighted his 8 keys of cloud transformation
1) The leadership will to invent and reinvent
2) Acknowledgement that you can't fight gravity
3) Talent that’s hungry to reinvent
4) Solving real customer problems with builders
5) Speed
6) Don’t “Complexify”
7) Use the platform with the broadest deepest set of tools
8) Pull everything together with aggressive top down goals
Except for the sales pitch on key #7, the other seven keys are leadership focused, not technology focused. The point that Jassy drives home at the start of the keynote is that culture and leadership matter the most when transforming a company to be cloud savvy.
To sum up Jassy’s keys to cloud transformation, it requires strong leadership that is willing to challenge current business models and industry norms, create a culture of people who love change, focus on using the cloud to solve real business problems not just pet projects, understand that speed is a competitive advantage, choose simplicity over complexity and drive the transformation with an aggressive future state vision that focuses on business outcomes. In other words, cloud for the sake of cloud just doesn’t cut the mustard.
On to the announcements. Jassy started by describing how AWS reinvented themselves in the areas of compute, containers, serverless, data, machine learning, hybrid cloud, and industry specific solutions.
Compute
Even though most compute in the public cloud is abstracted from the developers, there is still a need to innovate on performance, cost optimization and choice. And innovate they did! First up is Graviton2. This new generation of chip technology offers 40% improvements in price and performance. Inferentia is a new chip designed specifically for deep learning workloads while Trainium is a new chip set designed specifically for training machine learning models.
Containers and Serverless
As enterprises mature in their cloud journey and shift towards cloud native development, popularity of containers and serverless functions are soaring. But building and running containers, serverless functions and microservices is still way too complex. In alignment of key 6 – don’t complexify – AWS released a slew of new services and feature enhancements to tackle this problem.
After declaring that two-thirds of all containers running in the public cloud run on AWS, Jassy announced ECS Anywhere and EKS Anywhere (coming in 2021). Now AWS’s fully managed container services can be run in your datacenter. Jassy shared that 50% of customers were using serverless capabilities and announced that pricing would be changed to bill per the 1ms which could save customers up to 70%. He also announced Lambda container support so you can run your functions in containers and AWS Proton, a fully managed service for ease of deployment of microservices.
Data and Databases
Eating their own dog food on key #1, leadership must have the will to invent and reinvent, AWS took a hard look at their block storage services which launched back in 2008 and got a facelift in 2014 with the announcement of gp2. They announced gp3 which is faster more cost effective storage that separates IOPs from storage and allows you to scale IOPS independently. Jassy also announced Io2 Block Express which is the first SAN built for the cloud.
On the database front, Jassy announced Aurora Serverless V2 which can scale to hundreds of thousands of transactions per second and Babelfish for Aurora PostgreSQL which allows you to migrate from SQL Server with no code changes. One of the hardest parts of doing analytics is sourcing and moving data. To simplify this challenge, Jassy announced AWS Glue Elastic Views which allows you to set up materialized views of your data and manage all of the dependences for moving data from source to destination.
Machine Learning
I recently did a podcast on machine learning and one of the topics we discussed was that there are many ML experiments and proof of concept projects but very little actual production deployments. One of the reasons for that is it the complexity of training and deploying models. AWS took some bolds steps in addressing this complexity head on with several announcements.
First up was AWS SageMaker Data Wrangler which aims to simplify data preparation for machine learning. A client recently asked me for recommendations on how to share models amongst their peers both internally and externally within the industry. Low and behold, Jassy announced SageMakerFeature Store, a managed repository for machine learning models. To solve the challenges of deploying new machine learning models, AWS announced SageMaker Pipelines. Operations for machine learning is also very complex. AWS DevOps Guru was announced which uses ML under the covers to anticipate operations issues and recommend actions to remediate those issues.
Hybrid Cloud
Who needs the NIST definition of hybrid cloud when AWS has their own definition? Seriously though, Jassy said they took a step back and challenged what hybrid really means to them and here is what they came up with – “When you go back to asking what is hybrid, it’s not just cloud and on-premises datacenters. It’s cloud, along with various edge nodes, on-premises data centers being one of them.” AWS’s goal is to deliver the hybrid experience with a common set of APIs and tools no matter where the workload runs.
AWS announced that they would offer two new smaller sizes of AWS Outposts for use cases where there is less space such as hotels, restaurants, retail, etc. They now offer a 1U size, 40th the size of standard outpost or one pizza box size as Jassy put it and a 2U size (2 pizza boxes). In addition, they announced three new AWS local zones with 12 more coming in 2021. Jassy also discussed AWS Snowcone, a small rugged device that is used at edge locations where there is no or limited connectivity.
Then there was AWS Wavelength which provides 5g connectivity with the major telecom providers while abstracting the individual providers complex and proprietary integration and APIs to offer developers a standard interface no matter where you are connecting to.
Business Processes as a Service
The next category of announcements falls into a bucket I call business processes as a service. When we talk cloud, many people think of cloud as infrastructure or another datacenter in the sky. As enterprises start maturing in their cloud journey, they start moving up the stack towards platform or fully managed services. The cloud vendors in the last two years have been going up even higher in the stack and offering entire business processes as fully managed services.
AWS made several announcements around fully managed business processes. In my opinion, this is where the most value in the cloud is.
Since COVID sent most of us home to work remotely since March, I have seen a surge of client interest in improving or even outsourcing call center business processes. AWS previously released Amazon Connect, a cloud-based call center in 2017. They bolstered that offering with numerous announcements.
- Connect Wisdom - Uses ML to deliver faster insights in real time for agents. Includes built in connectors for Salesforce.com, ServiceNow and others
- Connect Profiles - connects to internal or 3rd party databases to return customer profiles
- Real Time Contact Lens - uses ML to detect customer experience issues during live calls
- Connect Tasks - workflow for call agent tasks
- Connect Voice ID - voice authentication to speed up validation process
In manufacturing, enterprises are embracing IoT and putting sensors on everything so they can monitor equipment and proactively address potential maintenance issues. But implementing these types of systems can be complex and time consuming. That is why AWS announced Monitron and Lookout for Equipment to provide an end to end monitoring and predictive maintenance solution as a service.
My Take
The rate at which AWS innovates is astounding. The number of new services and enhancements to existing services that they release each month is impressive and hard to keep up with. As they offer more fully managed services to help customers increase speed to market, they also continue to innovate at the infrastructure layer to drive down costs and increase performance.
As cloud providers make further investments in business processes as a service, it is becoming harder for individuals to continue to justify a cloud agnostic approach unless speed to market is not important to them. The effort and costs of building your own capabilities in the name of fear of vendor lock-in when the same capabilities are available as a fully managed service is a tradeoff with great consequences. As the cloud vendors move up the stack and deliver industry specific solutions as a service, those companies that embrace these managed services can gain a huge competitive advantage over those who build it themselves.
As AWS and the other major cloud providers continue to deliver higher level services, I predict most multi cloud strategies will shift from cloud agnostic workloads to best cloud for each workload. Workloads in the next 5 years will be more customized towards the preferred cloud provider and leverage higher level services to increase speed to market and offload the complexities of building and running the infrastructure and middleware to the cloud providers.
The faster the cloud providers can provide easy to consume fully managed business services that solve tough business problems, the more workloads they will win from enterprise customers. Expect a surge in industry specific managed services from the cloud providers over the next few years. For those companies that have moved beyond the misguided mindset of “the cloud is just someone else’s datacenter”, leveraging services at the business process layer will allow them to leapfrog their competition.
A special shout out goes to Mark Nunnikhoven for letting me use all of the screenshots he captured from the live event. Thanks Mark!