Serverless - State Of the Union

AWS DevDay: Nordics
Serverless: State of the Union
Ian Massingham | @IanMmmm
D e v e l o p e r/ Te c h n i c a l E v a n g e l i s m L e a d
A m a z o n W e b S e r v i c e s

A typical day for a developer…
Manager “We need an app to let our customers update preference
settings – our competitors just launched the same thing, so I need it
fast. I don’t want to pay a lot for it, especially when no one is using it.
But remember that we’re growing, so make sure it scales great and is
easy to manage and operate. And you’re on your own – sorry!”
Developer “Not a problem. I’ll make it serverless…”

© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Serverless means…
No Server Management Flexible Scaling
No Idle Capacity
$
High Availability

Joe promised his boss an app. 
Now what?
Step 1: Find a working example!
Serverless web app with
…an API hosted on Amazon API Gateway
…access to existing data stored in Amazon DynamoDB
…static content served by Amazon S3 and Amazon CloudFront
…dynamic content/business logic encoded as Lambda functions
What we need

© 2017, Amazon Web Services, Inc. or its Affiliates.All rights reserved.
• Search and browse ready-made apps and
samples
• Customize open source apps to get
started quickly
• Share apps privately or publicly
•Monetize APIs using theAWS
Marketplace
• Powered by AWS Serverless  
Application Model (AWS SAM)!
AWS Serverless Application Repository (preview)

Lambda Function Editor
• Edit multiple files at once!
• Create new files for existing packages
• Run tests and view results and logs
without leaving the editor
• Save test inputs and reuse them later
• WYSIWYG designer

Serverless app benefits
Go from concept to working code in under 5 minutes. And
your app isn’t just a prototype…
• 1,000 concurrent executions (10K TPS) by default.
• Every account in good standing is
automatically approved to 3X the default!

• 1,000 concurrent executions, pre-
approved to triple Faster perf with
cold-start
optimization! 
Up to 80% reduction in start
times for larger functions

Built-in logging and
monitoring, now with:
“grab-n-go” logs in the
Lambda console
approved to triple
• Up to 80% faster cold start times

1.5GB
3GB
Top memory size
of…
approved to triple
• Up to 80% faster cold start times
• Find Lambda logs instantly

New runtimes GA from Jan 15
.NET Core 2.0 Golang

Next, Joe needs to make some changes.
Step 2: Customize and share
Need an IDE for local dev/test/debug
Customize the app we deployed in step 1
Set up a CI/CD pipeline for the team
Deploy (safely!)
What we need 
to do

Easy-to-use IDE in the cloud
Builtin GitHub support
Builtin Lambda blueprint support
Builtin SAM Local:
Local testing!
IDE debugging!
Deploy directly to Lambda and/or push
updates to GitHub
AWS Cloud9

Automated CI/CD process
Git push your changes from Cloud9
when ready to share…
Set up an AWS CodePipeline to build
automatically on updates
Local Testing

I need to deploy Lambda functions
safely and incrementally.
Customer feedback
How can I deploy multiple
functions and APIs in a
coordinated way?
How do I roll back
automatically?
How can I shape traffic
between two versions of
an API?

Safe, incremental serverless deployment
Lambda: Weighted aliases
• Set percentage of traffic between two versions of the same
function
API Gateway: Sub-stages for incremental deployments
• Share traffic between two stages
AWS CodeDeploy: Automated support for safe serverless
deployments
• Amazon CloudWatch metrics-driven rollout/rollback

Another typical day…
Manager: “That app you built is working great, but now we need to
include data from a relational DB and one of our legacy services in our
VPC. Is your serverless architecture going to handle that?”
Developer “Not a problem.”

Next up: Deal with legacy stuff.
Step 3: Connect to enterprise systems
Monitor concurrency
Limit concurrency for legacy connectivity
Hook up APIs to VPC-based services
Audit and log all invocations
What we need 
to do

I need a way to protect the capacity of
a Lambda function from others running
in the same account.
Customer feedback
I want to temporarily
disable a function.
Can you create a “dev mode”
for functions?
Lambda can overwhelm
my legacy systems with
too many calls

I need a way to protect the capacity of
a Lambda function from others running
in the same account.
Customer Feedback
I want to temporarily
disable a function.
Can you create a “dev mode”
for functions?
Lambda can overwhelm
my legacy systems with
too many calls
Insight: Concurrency control is a key
element of managing functions in a
serverless app.

Lambda: Concurrency metric
Lambda: Per-function concurrency throttles
Enterprise connectivity
What you can do with it
1. Limit concurrency to legacy systems
2. Protect serverless production services from
other functions in the same account
3. Temporarily disable a function
4. Develop functions with limited billing and
“runaway” protection

VPC (PrivateLink) integrations
Access services and data in
your VPC from an API
Regional endpoints
Fast, region-specific API
endpoints
Enterprise connectivity: API Gateway

API Gateway connectivity solutions
To
From
Internet
Regional 
AWS Service
Customer VPC
Internet
Regional 
AWS Service
Existing Service 
(Edge-optimized, public integrations)
Private VPC
Region-optimized Endpoints
Region Endpoints
+Private VPC
Recent
New

Log and audit
Structured logging for APIs
AWS CloudTrail support for
Lambda functions

Serverless for the enterprise: checklist
Uptime
Scale
Performance
Management
Safe deploys
Concurrency controls

Thank You
Ian Massingham | @IanMmmm
D e v e l o p e r/ Te c h n i c a l E v a n g e l i s m L e a d
A m a z o n W e b S e r v i c e s

Serverless - State Of the Union

Recommended

Recommended

More Related Content

What's hot

What's hot (19)

Similar to Serverless - State Of the Union

Similar to Serverless - State Of the Union (20)

More from Amazon Web Services

More from Amazon Web Services (20)

Serverless - State Of the Union