Top 5 lessons learned at AWS Re:invent

4 Comments

Credit: Amazon Web Services

Now that the smoke is clearing (and echoes of Skrillex are fading) it’s time for a quick post mortem of the third annual AWS Re:Invent in Las Vegas.

Cloud is not all about price anymore.

What’s this now? An Amazon Web Services event without any price cuts? Unheard of. That could be because AWS is pushing to prove that its cloud is about way more than low-level pipes and gears. To be fair, it never was really all about price, but face it, price cuts on basic compute and storage by AWS, [company]Google[/company] and [company]Microsoft[/company] have driven much of the dialogue around public cloud computing.

Andy Jassy, Amazon’s SVP of web services, repeatedly invoked agility, not price, as a prime motivator for cloud. But Amazon’s shift could also be due to the fact that rivals Google and Microsoft have more cash and may be better able to weather price cuts, so AWS needs to change the subject.

Plus, there are price cuts and there are hidden price cuts. The new Lambda service could lower prices by eliminating the need to run a whole stack of services for bursty, periodic or intermittent workloads.  (The Lambda free tier covers up to 1 million requests per month and 400,000 GB seconds of compute time per month.) So it’s a price cut in another form, I guess.

Both Werner Vogels, Amazon’s CTO and Terry Wise, director of worldwide partner ecosystem (pictured below), talked about a variety of AWS and partner services as forming the basis for richer applications. And the company trotted out partners Docker, Splunk, Pristine, [company]Omnifone[/company] and others to illustrate that. As IDC analyst Larry Carvahlo pointed out, big legacy IT companies that continue to sniff that AWS is “just” an infrastructure company are either delusional or lying. (Well, he didn’t say lying, he’s far too nice for that, but you get the drift.)  The lines between IaaS and PaaS are blurring more every day.

Terry Wise, director of AWS Worldwide Partner Ecosystem  speaks at the APN Summit at AWS Re:Invent.

Terry Wise, director of AWS Worldwide Partner Ecosystem, speaks at AWS Re:Invent.

 

Amazon wants to manage ALL your IT

The upcoming AWS Service Catalog will help customers manage their AWS resources. That means that you, as an authorized user, can access Service Y (and it will be the correct, updated version of Service Y), but an unauthorized user cannot.

As evangelist Jeff Barr blogged:

It is a tool that will allow any IT department to deliver AWS-powered services to internal users while maintaining consistency and control.

But if past is prelude, that catalog will soon cover services running outside of AWS in other clouds and potentially on premises as well. Note that AWS just launched an AWS Directory Services that ties into on-prem Windows applications and a portal capability for VMware admins.

“I can guarantee you that this service catalog will include access to other cloud vendors very soon,” said Cloudyn CEO Sharon Wagner. “AWS wants IT managers to use its management platform and not others.” In Wagner’s view, he who owns the service catalog soon owns provisioning.

EC2 Container Management Service is nice, but it’s no Kubernetes.

Amazon’s new EC2 Container Management Services — the long-awaited Docker announcement — is now in preview. Per the AWS blog:

This service will make it easy for you to run any number of Docker containers across a managed cluster of Amazon Elastic Compute Cloud(EC2) instances using powerful APIs and other tools. You do not have to install cluster management software, purchase and maintain the cluster hardware, or match your hardware inventory to your software needs when you use ECS.

From what he can tell from early documentation, this is not a true competitor to Kubernetes-based Google Container Engine, said Sebastian Stadil, CEO of Scalr, a cloud management company.

“Amazon’s product does not manage the underlying VMs, while Google Container Engine does,” he said via email. If that is the case, he’s not sure why companies would run this and not just deploy Kubernetes on EC2.

In response, an AWS spokeswoman said the company’s customers wanted a service that lets them run their applications on their own Amazon EC2 instances without worrying about scaling. This approach gives them the flexibility to choose their existing Reserved Instances, use Auto Scaling groups, etc.

The spokeswoman later added:

“With other container management solutions on EC2 you need to run cluster and configuration management software, manage their scaling and availability and plan for the kind of resources you will need for those. With Amazon EC2 Container Service, you can launch as many clusters as you like at any scale without the need to run any cluster or configuration management software.”

 

Conference Wi-Fi still stymies AWS

If any company should be able to nuke the Wi-Fi black hole that afflicts all large conference venues, you’d think it would be Amazon, and yet … reporters in the keynote hall were cut off. I know provisioning Wi-Fi for 13,000 always-connected people in a single venue is tough, but isn’t this the sort of scale problem AWS was born to solve?

If you needed to file (or tweet or whatever) the keynote, you were better off watching the live stream from a remote location — which as one of my weary colleagues pointed out, begs the question why you even go to the event in the first place. So, AWS (and every other tech vendor out there), we’re begging you: Figure this out.

Reinvent-ticker

Note: This story was updated at 7:10 p.m. PST to include the AWS statement regarding EC2 Container Service and again at 11:50 a.m. PST on November 18 with additional AWS comment.

4 Comments

David Mytton

Aside from the container announcement, the key is that products focused on pure compute+storage+networking are done. These things are commodities and barring major changes to architecture (such as a possible shift from VMs to containers as the unit piece) and performance/cost improvements, all the big, interesting announcements are the supporting services. Lambda. CodeDeploy. Service Catalog.

AWS is building like a startup – using the core IaaS infrastructure (EC2, S3, DynamoDB) to build SaaS.

Comments are closed.