itb2016 - aws lambda

AWS LambdaWhat, why, when, how?

What is Lambda?

Compute service where you

don't think about…

Servers

Capacity

Deployments

Scaling

Fault tolerance

Metrics and logging

Automatic OS and

language updates

All of that is taken care of

automatically

You can easily…

Bring your own code,

including native libraries

Run code in parallel

Lambda use cases

Create backends

Create event handlers

Create data

processing systems

Serverless

Event-driven scale

100ms billing units

You never pay for idle!

How it works

Upload code in a zip file

Code can be…

v0.10 & v4.3 Java 8 v2.7

Code is stateless,

persistence goes elsewhere

Monitoring and logging

handled automatically by

CloudWatch

Integrates with other

AWS services

S3

DynamoDB

RDS

VPC

CloudWatch

CloudFormation

Cognito SNSAPI Gateway

SES

SWF

Kinesis CloudTrail

CloudWatch

All these & more…

Compute Power

23 "power levels"

Specified in Lambda

as memory

From 128MB

To 1.5GB

More memory = More CPU

Price scales with

power level

Pricing per 100ms(rounded up to nearest 100ms)

Run durations from

100ms to 5 minutes

For compute intensive tasks

set power level higher

For I/O intensive tasks

set power level lower

Event Sources

• S3• Schedule events• AWS IoT• CloudWatch

• DynamoDB• Kinesis• SNS• API Gateway

(sort of)

“Serverless” API’s with

AWS API Gateway & Lambda

Versioning

By using versioning

nothing changes

But now you can

publish a version

Versions can have aliases

VPC Support

Pricing

$0.20 per 1 million requests

Pay for CPU usages in 100ms

intervals. Price depends on

memory allocation.

From $0.000000208 to $0.000002501 per 100ms

First 1 million request

and

400,000 GB-seconds of

compute time per month

FREE!!!

This is equal to approx.:

888 hours at 128Mb

….

74 hours at 1.5Gb

Example uses

Image processing

Data processing

Scheduled Tasks

Document generation

If you need help email me

andrew.dixon@mso.net