Pivotal Paris 2019: Cloud-Native Track Serverless, oui mais pour quels usages ?

1. © Copyright 2019 Pivotal Software, Inc. All rights Reserved.
Pierre-Antoine Porte, Platform Architect, Pivotal
4 juillet 2019
Serverless, oui mais pour
quels usages ?

2. Serverless ?

4. Serverless
“Managed services that scale to zero.”
Rachel Stephens

5. Is PaaS serverless ?

6. Platform as a Service
Platform-based service is a category of cloud
computing services that allows developers to
manage applications without the complexity of
building and maintaining the infrastructure.

7. Abstractions
Hardware
IaaS
Container Orchestrator
Application Platform
Serverless
Functions
Higher flexibility and
less enforcement of
standards
Lower development
complexity and higher
operational efficiency

8. XaaSYes, this name really exists

9. Yes it’s FaaS...

10. Scale to zero
g

11. Scale to zero
g g gg

12. Apps are now running on an infinite pool of resources
Pool of Resources
Thanks to this level of
abstraction, you can
scale any apps from 0
to +∞

13. Write less code

14. Write less code
@SpringBootApplication
public class Application {
public static void main(String[] args) {
SpringApplication .run(Application .class, args);
}
@Bean
public Function<Flux<String>, Flux<String>> uppercase() {
return flux -> flux.map(value -> value.toUpperCase ());
}
}

17. Now who’s really using it?

18. Don’t do what does not work

19. Don’t do what does not work
Let’s not make the same
mistake some people made
with containers.

20. Don’t do what does not work

21. Don’t do what does not work
Please.
Don’t break your APIs into
functions just because the tech
is cool.

22. Event-based
Integration
• Scheduled tasks, data
integration
• File processing
e.g. images and videos
• Security scanning
• Complex Event Processing
and Change Data Capture
• Monitoring, notifications
and alerting
• Custom auth
e.g. via API Gateway
Web Events Stream
Processing
• Website back-end services
like form post handlers,
authentication, tracking and
logging.
• APIs to back-end data
services for mobile and
web apps e.g GraphQL
• Webhook handlers
• Chat integrations
• Digital assistant services
e.g. Alexa skills
• IoT streams
• Log ingestion
• Event streams e.g. with
Kinesis
• ML pipelines
Use Cases & Scenarios for functions

23. Tech is cool - Function start is even cooler
There’s no magic.
If you don’t have something
constantly running, then the
first time it runs it should also
be loaded.
That’s what the industry calls:
Cold Start

24. There’s no such thing as OCI (open container initiative) for serverless.
Open Source is a good choice. Widely adopted technology is even a better one.
Be careful with lack of standards
And that’s why we chose to build our FaaS abstraction with KNative, an open source
framework on top of Kubernetes.
We also created an open source project, riff, which build and run function integrating Knative.

25. Some real use-cases
Most famous company doing streaming
Separating videos in 5 mins chunks and encoding it and reassembling it
Media and Information company
Real-time analytics of product offerings to enhance UX
Famous US newspaper
Image resizing for mobile, tablet and desktop views
5 min chunks Encoded video
Receive video from
the producer Full video

26. Execute functions in response to
events. Use PFS to handle web
events, event-based integration,
and large scale streaming data.
Pivotal Function Service (PFS): A Runtime for Functions
Trigger functions via HTTP/Message Broker — With PFS, you can
quickly set up functions for a variety of events. PFS is architected to
support pub/sub message delivery over messaging channels.
Run functions anywhere — PFS lets you easily run functions
on-premises and in the public cloud for maximum flexibility.
Built on Knative — PFS is built on Knative, an open-source project led
by Google. Knative simplifies how developers deploy functions atop
Kubernetes and Istio.
Pluggable build system — PFS features a source-to-container
mechanism to simplify deployment. Use proven components like Cloud
Foundry Buildpacks.
Pluggable event sources — PFS Event Sources facilitate the creation
of Feeds from a variety of external event sources such GitHub
webhooks, blob stores, and database services.
Pluggable event brokers — PFS can be connected easily with popular
message brokers such as Kafka, Google Pub/Sub, and RabbitMQ.

28. Transforming how the world builds software
© Copyright 2019 Pivotal Software, Inc. All rights Reserved.