Thursday, November 13, 2014

AWS Re:Invent Day 2 Keynote Live Blog

Got here early enough to get coffee and get a good seat! Crap, seat not so good, still can’t see slides well… *sigh*

Werner Vogels on stage - Quick recap of yesterday’s announcements. Says the party artist will be announced at the end of the keynote

Talking about building applications on cloud - Harder, Better, Faster, Stronger than they have ever been. (hint hint)

Services delivered in a broad ecosystem make the difference (trying really to differentiate on services it would seem) vs. just an IaaS platform

Splunk on stage - All core products run on AWS, Splunk cloud (they run it for you), Splunk Enterprise, etc.

What has changed over the last year? Customers are moving from just dev/test and peak apps and moving true production workloads en masse to AWS. Splunk can help with visibility between on-prem and AWS.

Mentions customers - Coca-Cola, Nike and their use cases. often POC on Amazon and then moving production to AWS. Saved time and money using AWS.

Mention of Finra - stock trading security regulator - no more standing up hardware, they moved all applications and Splunk to AWS to focus on what matters, not management of infrastructure. Mention of multiple regions and APIs for scalability 

(I notice almost all guests on stages mention that, must be in the speaker notes for everyone. AWS is hitting scalability, API’s, and services as differentiators)

Werner back on stage - Slide -> AWS is Secure, Adaptive, Resilient, and Global. talking about “pushing a button” to make infrastructure appear

The Application Extends the Platform - talking about importance of API’s and extension of the platform in infrastructure as code and fitting tis into emerging application development models

(As an aside, the Splunk dude that just spoke sat down next to me… awkward)

OmniFone on stage - online music platform, talking about the music industry and the complications of music as an industry. They started with a 15 million platform, it didn’t hold up to the load. They could’t iterate fast enough. They had to start over and started over on AWS. “AWS was the only choice” (Also noticing that as a common theme of the guests, they are all saying it).

Now has a geographically scalable, redundant services across the globe on AWS. Building this platform has allowed the music industry to build what matters. They have delivered more audio/video faster than ever before.

Talking about high res quality sound and the challenges (about 150 times the file size of typical mobile file delivery). How do you deliver the large files in a large uninterrupted stream? talking about Podio (Neil Young’s company?) and what they are doing there

Werner back on stage - Broad Services drive the speed of development, talking about “agility as the Holy Grail” of application delivery. Increasing consumer choice is driving the market to a new model that needs to be agile and fast. Dev & Test is the Core to Agility

Says today budget’s of most CIO’s for Dev & Test are between 40%-60%. How do you optimize that and make that portion of the budget faster.

The Weather Company (The Weather Channel) on stage - talking about weather as a science and data platform. How do you great services based on information you can’t control but potentially affects both business and lives all around the world.

They have built a platform on AWS to feed others (Apple, Google, Yahoo, etc.) to move beyond cable. Also feed data to all major airplanes to help with traffic control. Provide data to local broadcasting companies all over the world. They want to be the “data warehouse” for all things weather in the world.

They didn’t start this way, had a traditional model of physical data centers with physical hardware. They had to change both the infrastructure as well as the culture. 

(I like they brought the human aspect into this, not just technology, so often overlooked)

They choose AWS for scale (scalability point hit again), as well as confidence in the services. Platform has provided close to 100% uptime and weather forecast is less than a second by analyzing over 800 sources around the world. The platform allowed them to “go faster” and constantly improve the accuracy of forecasting over time.

Over 1 billion devices served from the platform between IOS8, Android, and downloads of apps on Mac/PC

Werner back on stage - Development is changing to support agility

Pristine (Google Glass specific company with a focus on healthcare) on stage - They are using AWS and…. drumroll please…. Docker!! (You knew it was coming!)

Slide - Containers are the key to our growth, this allows them to develop once and run everywhere. Rollback are simple, etc.

Talking about the combination of AMI’s for the base image and the layering of containers on top is the “perfect match” for them and allows them to go as fast as possible and scale beyond anything else that is out there.

Werner back on stage - Why do developers love containers? Going into to all the usual containers value proposition. talking about containers do present some overhead challenges set up.

Announcement - AWS Container Service - deploy environment to make containers easy. (huge applause). All with an API, integrates with Docker repositories, also integrates with Mesos

Demo of Docker containers into the system on stage now (I can’t see the screen well sadly). Instances (AMI’s set up), register the cluster with the service. Name the Docker image that will be used, start running the task. Single instance, deploy and scale to 5 instances, deploy front end.

Scale up to 30 instances (different instance types as well)

(Got a call.. had to step out… I’m sure it was awesome… sorry about that)

Docker CEO, Ben on stage - Where isn’t Ben these days?! Good for him and good for Docker….

Developers are content creators - Docker removes the “crap work (my words)” from development and allows developers to go faster.

5 steps to containers - 
1. isolation of process in an OS
2. good API’s to run anywhere
3. create an ecosystem (Docker Hub)
4. create a new container based app model
5. create a platform for managing it all

Talking about Gilt.com - joint AWS and Docker customer, before docker 7 apps and hard to deploy, 300 micro services and 100+ deploys per day

Just passed over 50 million downloads of Docker!

Werner back on stage - Simplification drives reliability and performance

What are the primitives of cloud in an execution environment?

talking about data, triggers, and actions of applications. A data change triggers an action to update other portions of data.

Why don’t we architect that way? need to create a full, complex stack to “run a function and modify data”

Announcement - AWS Lamba - An event driven computing service for dynamic applications. You just write code and no underlying infrastructure (it’s always there somewhere, they are just taking it away so you don’t have to worry about it)

Basically state changes and events drive the system (new pricing model?) - write code without infrastructure.  - (Another PaaS without calling it a PaaS?)

code only runs when needed - cost efficient and efficient

Really interesting concept - Talking about IoT (Internet of Things) and triggers as the new currency

Netflix on stage - talking about micro-derives and Lamba, they can replace inefficient existing services with trigger based serves.

Encoding Media Files is an example - get file from studio, chunk it up, process it, ship out to CDN’s
Backup for Disaster Recovery - they can now do backups based on triggers and events vs. time
Security - when an instance is spun up, trigger security check to make sure it is configured correctly

Werner back on stage - Units of cost for Lamba - number of requests, execution time - there is a free tier for each customers each month - today it is available as a preview.

Announcement - New Instance offering C4 (based on Haswell processor), up to 36 vCPU’s, EBS optimized by default and included in the price.

Announcement - New EBS - SSD backed EBS up to 10,000 IOPS (up to 160Mps) and 20,000 IOPS (up to 320Mbps)

Intel on stage now - talking about C4 instance… speeds… feeds… The processor is actually an AWS exclusive

My take: It would appear they have hit on a few key differentiators to move forward beyond iaaS. Scalability (to differentiate from on-prem), API’s for developers (to differentiate from other public clouds), and services across the broad ecosystem. They want to be the developers model of choice and seem to get the only way to get “next generation applications” is to enable the developers and start down the micro-services and containers path.

Well played AWS… well played…

Over all, super impressed with year vs the keynotes of past years.

No comments: