What I learned from AWS re:Invent 2018

A few quick thoughts as the world’s biggest cloud conference comes to a close.

Cloud consulting jobs will get harder. Mid-sized firms should be happy.

There were a number of announcements that will make it more difficult for cloud architects, consultancies and IT shops operating for customers in the AWS cloud.

Between the Well-Architected Tool that can check a workload against best practices to Security Hub and Control Hub, AWS is yet again handling the undifferentiated heavy lifting, this time the rote consulting work that pads out cloud implementations.

This is great news for customers and should make it easier for mid-sized companies to start cloud migrations with greater confidence. There is a huge gap in the marketplace between startups in AWS (practically all of them), enterprise in AWS (almost all have some projects there) and mid-sized businesses, which have almost ignored the cloud’s existence.

Many mid-sized firms simply lack the resources to migrate and are waiting for their depreciation cycles to complete, so any barrier tends to be fatal to cloud adoption. This gives those firms some superpowers to get the process moving without getting sucked dry by consulting partners.

Many developers don’t understand TCO.

Jared Short nailed this:

Hacker News tends to be a place where a bunch of generally smart people say some very stupid things. But Jared’s comment reveals something we see in development groups all the time and it’s highlighted by comments about pricing in managed services. Put simply, unless your developers work for minimum wage, your Total Cost of Ownership calculation is wrong.

Cloud has been around long enough now that the community needs to start pricing this correctly, including personnel, data centers and utilities, and stop skewing the calculation to support the status quo.

Serverless is going after the mainframe.

If you’ve ever been around a mainframe retirement project, you’ll know it doesn’t retire and neither do the 70-year old COBOL developers still stuck in maintenance hell. Mainframes are painted as monoliths but Ryan Marsh’s tweet is key — they are not.

Bringing your own language to Lambda is a great feature and will be horribly abused by some people to build containers. But the sub-text of this launch is that you can run COBOL and start hacking away at mainframe jobs. It’s the only way to avoid the all-or-nothing approach to mainframe migration that rarely works. It’s also the only hope these poor COBOL devs for retirement.

Are you really focusing on your customer?

One of the big reminders every year at re:Invent is just how Amazon is nothing if not customer-obsessed. Everything is built for customers, from their feedback, their problems and comments. Everything.

It’s the reason why Amazon extends its lead as clearly the most innovative company in the world, but it’s based on a principle that all of us should follow. When we develop pet projects, products to chase competitors or ideas that are built on internal company politics, we betray our customers.

Tech aside, we should all be copying this guiding principle from Amazon. Listen to your customers, deliver what they want and good things will happen.

Serverless is mainstream.

It’s been a lonely year for those of us who adopted serverless early — truly, it’s been the frontier of cloud and it can be a difficult place to be. AWS has rewarded the Serverless Faithful this year with a massive list of feature releases that address almost all the major limitations of this paradigm.

I don’t know anyone in the serverless world who isn’t ecstatic with the focus on serverless tech, and AWS’ clear commitment to all things serverless going forward.

Final observations

This is the smoothest re:Invent I’ve ever experienced — it ran perfectly despite the record attendance. The session and workshop catalog provided options for everyone and there was more than enough content to justify the 4–5 day length. Vegas continues to be an excellent location for this conference.

The Keynotes this year delivered an astounding number of very significant launches that dwarf anything I’ve ever seen from Azure or Google. AWS was already well ahead of the others but have extended the lead into lightyears. It’s no longer possibly to compare clouds against AWS — finally, most of the offerings simply have no counterpart.

It’s been an exhausting and exciting week but already I’ve blocked off my calendar for re:Invent 2019.