DockerCon 2019: Opensource Summit Serverless

Serverless is complementary technology running side-by-side containerization. In this session, using OpenFaas and Nuclio, panelists will present use cases via a series of demos and Q&A followed by birds of feather round table discussion. This session will go beyond hello world and highlight on-prem usage of serverless and containerization technologies and the real use cases in the enterprise.
Hosted by Rupak Ganguly - Solutions Architect, Docker with panelists:
Yaron Haviv - CTO, Iguazio | Burton Rheutan - DevOps Lead, OpenFaaS Core Contributor | Pav Jimanov - CEO, SoftwareDefined

Transcript

1. RUPAK GANGULY Solutions Architect, Docker OSS Summit: Serverless @rupakg
2. PAV JIMANOV Principal, SoftwareDefined Panelists: OpenFaaS BURTON RHEUTAN DevOps Lead, OpenFaaS Core Contributor @PavJimanov @_burtonr
3. YARON HAVIV CTO, Iguazio @yaronhaviv Panelist: Iguazio
4. Agenda Overview OpenFaaS • Presentation • Demo Iguazio • Presentation • Demo Q&A Roundtable discussion
5. really… serverless?
6. “Serverless usually refers to an architectural pattern where the server side logic is run in stateless compute containers that are event-triggered, ephemeral and fully managed by third party.” - Martin Fowler of ThoughtWorks serverless?
7. serverless and containers are complementary technologies Serverless vs. Containers
8. OpenFaaS Burton Rheutan @_burtonr OpenFaaS Core Contributor 118 118 Money DevOps Lead Pav Jimanov @PavJimanov 2019 Dockercon Principal at SoftwareDefined Docker Captain
9. The OpenFaaS Story • Began Nov ‘16 • DockerCon Hack by Alex Ellis • Containers for Alexa on Swarm • Added to CNCF Landscape • 200+ Contributors • 29+ Production users • 2 Bossie Awards • 2017 & 2018
10. https://kenfdev.o6s.io/contributors-page#/
11. Core Values Operator Friendly Developer First Community Centric
12. Function Templates ...and more! Template Store
13. OpenFaaS Function $ faas new http-status –-lang python requests def handler(req): r = requests.get(req) return str(r.status_code) handler.py requirements.txt Python Template Docker Image Container Registry
14. Event Connector Kafka OpenFaaS Gateway Function 2 payment.created Kafka Connector invoke Function Topic Function 1 cart.checkout Function 2 payment.created cart.checkout payment.created Function 1 cart.checkout
15. Event Connector Kafka OpenFaaS Gateway Function 2 payment.created Kafka Connector invoke Function Topic Function 1 cart.checkout Function 2 payment.created cart.checkout payment.created Function 1 cart.checkout • CloudEvents • Kafka • HTTP • Cron • AWS SNS • Redis • MQTT • NATS • ...and more
16. Serverless...on servers Image courtesy of Chris Watterson
17. Serverless...on servers • Portable • Docker Swarm • Kubernetes • Hashicorp Nomad • AWS Fargate • Raspberry Pi (ARM) • Lightweight • Run in existing cluster • Separate namespaces (K8s) Image courtesy of Chris Watterson
18. Architecture – Kubernetes native Gateway NATS Prometheus OpenFaaS Operator Function 1 Function 2 Ingress/LB Alert Manager CRD
19. Production Users
20. • Join the Community • https://docs.openfaas.com/community • Try out the workshop • https://github.com/openfaas/workshop • Support the project • https://opencollective.com/openfaas What’s Next
21. Thank You!
22. Local Operation with Swarm Function OpenFaaS Gateway VueJS UI C# API default network func_functions network
23. Portability Demo OpenFaaS FaaS-netes Public Cloud Kubernetes FaaS-Lambda OpenFaaS AWS Lambda Docker EE Kubernetes FaaS-CLI
24. YARON HAVIV CTO, iguazio Nuclio: Serverless for data and compute intensive tasks @yaronhaviv
25. Main Challenge Today: How to Innovate & Iterate Faster Complex, Long Development ƒ Many moving parts, lack of skilled developers, scientists, and DevOps Production readiness ƒ Error-handling, security, logging, auto- scaling, live updates, .. Slow to respond or act ƒ High-latency, or fast but inaccurate ƒ Cloud latency/bandwidth barriers Modern Apps Today: Serverless Automate dev and ops High-performance Data, AI and app services Federated multi-cloud, on-prem or edge The Need:
26. But, Serverless Comes With Challenges Slow performance, lack of concurrency Stateless, limited application patterns Limited number or cloud specific event sources Hard to debug and diagnose, build dependencies Cloud vendor API lock-in
27. ƒ Non-blocking, parallel ƒ Zero copy, buffer reuse ƒ Up to 400K events/sec/proc ƒ GPU Support Nuclio: Taking Serverless to The Next Level Function Workers Event Listeners Open-source Serverless for compute and data intensive tasks, 100x faster than AWS Lambda ! Extreme Performance Shard 1 Workers Workers Shard 2 Shard 3 Shard 4 Workers Advanced Data & AI Features DB, MQ, File Functions ƒ Auto-rebalance, checkpoints ƒ Any source: Kafka, NATS, Kinesis, event- hub, iguazio, pub/sub, RabbitMQ, Cron, .. ƒ Jupyter, Spark, Rapids integration ƒ Data bindings ƒ Shared volumes ƒ Context cache Statefulness nuclio processor
29. 5 Best Match: Serverless + Kubernetes DBaaS Object (S3) • Cloud independent APIs, on-prem, edge • Auto-scaling + abstract infrastructure • One platform for serverless & containers • Abstraction: Auto generated code, Docker files & YAMLs • Automated dev & ops flow, observability • Maximizing performance and resource efficiency • Everything is a Kubernetes resource • “Serverless” data services • Endless scalability
30. Demo: Build & Test an App in 2 min
31. 7 Main usage: Fast, Intelligent, Data Driven Apps Persistent Data and Computation Cluster Scalable & Open ML Pipeline Serverless Models & APIs Automated Collection & Pre-processing Monitor Intelligent & Interactive Apps Simple and self-service consumption Real-time and historical data Open Services Data Sources
32. 8 Demo: Real-Time Analysis of Financial Data RT Tweet Sentiment Analysis Tick feed Analysis & Tagging Real-time Dashboard News Stream viewer World Trading Data Data Exploration & Analysis • Enriched tweet stream • Stocks tables • Stocks + sentiment TSDB
33. 9 Auto-Healing Network Operations ƒ Replaced a complex Hadoop based data pipeline that was never productized ƒ Cross correlating real time data from multiple sources with historical data ƒ AI based predictions trigger pre-programmed actions that fix evolving problems in the network ƒ Implemented within weeks Predict and avoid network outages in real-time
34. 10 Demo: Predictive Infrastructure Monitoring NLP processing Of real-time logs Network data Exploration & Correlation ML Training, Model export Failure & Anomaly prediction Real-time telemetry Auto-deploy Real-time DB
35. 11 ƒ Eliminating scale, operations and cost challenges ƒ Nuclio comes with a bunch of advantages: o Make your apps faster and simpler o Support many data sources (not locked to cloud) o Comes in open source or fully managed flavors Serverless Is rapid dev & ops for microservices Summary My Email: yaronh@Iguazio.com
36. Roundtable Discussion
37. Rate & Share Rate this session in the DockerCon App Follow: @rupakg, @pavjimanov @_burtonr @yaronhaviv Tweet: #DockerCon