Andy Jassy took to the keynote stage for the first official AWS re:Invent keynote of 2016, and it was exactly as many expected. The energy in the room was exciting, and the announcements were fairly well received across the board. When we are talking about why AWS matters to consumers, it was summed up rather nicely by the team at Fortune:
Agility over cost was the message from Andy. Customers of AWS predominantly choose to attack the challenges of agility first and price second. This is something that I've definitely seen in the market, and clearly forgot to mention the second part of that statement which should be "but they realize quickly that agility comes at a cost". That's a whole other story unto itself.
AWS Sells to People, and also to People who Sell to People
Logos were all over the Systems Integrator and ISV partner slide. This is where AWS has really seen a boom in their market share, which is being helped by others selling solutions atop the AWS infrastructure. AWS has been strongly pushing to grow their own offerings and enable the partner ecosystem as a result.
Companies running SaaS offerings also lead out the top consumers of AWS services in order to pass through the goodness to their customers which is a win for all. There can be little doubt about the success of the model which Jeff Bezos has built and evolved in the recent years as AWS has become the de facto standard in public cloud.
Three New Services a Day
This year alone, there have been approximately 1000 services launched. This can range from existing services being expanded into new regions, new availability zones, new services in general, and more. That translates to a rather powerful metric if you think of it.
"Every day that you wake up, you have an average of 3 new services that you can choose to use" - Andy Jassy, AWS re:Invent Keynote, November 30
Speaking of new services, there were a few things launched or announced as preview with a pending launch in Q1 2017. On the EC2 Instance offerings, we saw the addition of a few new members of the EC2 families:
- T2 gets 2 more sizes, XL and 2XL for more memory
- R3 (memory-intensive family) gains another family member in R4 with twice as much memory DDR4 for twice as fast and more caching
- I2 (I/O Intensive family) adds I3, adding 9x IOPS, 2x storage, 2x memory, and 2x vCPU
- C4 (Compute-optimized) welcomes C5 at 2x vCPU (Skylake) for 2x performance, 3x throughput to EBS, and 2.4x the memory available
GPU gets some love with the Elastic GPUs for EC2 where you can add any GPU elasticity to existing EC2 instances. This allows you to use the same code on your instance and have the GPU offload the processing. You were required to use the specific GPU accelerated instances with limited flavor offerings in the past.
Amazon Lightsail - VPS Made Easy
As easy as AWS makes it to spin up EC2 instances, they decided to extend the automation a bit further with the new Amazon Lightsail offering. You pick what you want, and all of the underlying infrastructure setup is done rather than you clicking through the wizard.
Instance and app images are available in a variety of product offerings today with more to come as we see the new service evolve:
The Amazon Lightsail is available at 5$ per month and up, with a many different levels of instance plans:
Lightsail has a rather familiar look for fans of Digital Ocean and other VPS cloud platforms. They do say that imitation is the best form of flattery. Unfortunately for those other providers, this opens the door to much more competition from the Amazon juggernaut.
For hardware acceleration, F1 instances using FPGA allow you to build and write your own accelerations to deploy, or you can use the AWS toolkit to upload to the F1 instance.
Learning Inside, Sharing Outside
"We always try to take what we use for ourselves to make it available to you" Andy Jassy
The F1 instance for FPGA acceleration is available in preview with full launch coming soon.
So, today they have announced the following products on top of the general directional updates that have been provided:
- 4 new instance families
- Added GPU
- VPS made simple
- FPGA on-demand
"You could build a house if you wanted to with a hammer and wrench and nails [...] it probably won't look very nice. The same is true with product development. It's so much easier to do when you have the right tools available to you" Andy Jassy
AI for the Masses
Data is powerful when you have an understanding of the data itself. Using the AI, Deep Learning, MXNet and Amazon Machine Learning platforms gives a variety of options. The goal is simple: More AI for the masses without having to try to build those services themselves. A new set of AI services are launching with the first 3 being as follows:
Amazon Rekognition - object recognition in images including face recognition with sentiment and face-matching
Amazon Polly - Text-to-Speech service - Text in, MP3 out. Not just audio, but life-like audio in multiple voices
Amazon LEX - Natural language understanding powered by Deep Learning that drives ALEXA
Those are just three specific new products being added to the ecosystem among many other improvements, and broader addition of services and features within the existing environments.
Quick Summary Notes
AWS does something amazingly well: create products to use inside, hone them, share them outside. In the same way that Kubernetes has hit the ground running because of 10 years of internal innovation, AWS products are the very same products that have been powering the Amazon infrastructure already. They just open the doors when it is ready, which enables us to consume the products as services, ultimately making them even more powerful and understood to grow the products even further.