Share
Embed
DevOpsDays Victoria 2019: Microservices and Databases with Kubernetes
DevOpsDays Victoria 2019
Chris Van Tuin
May 31, 2019
Transcript
86 slides
-
A DEVOPS STATE OF MIND: MICROSERVICES AND DATABASES WITH KUBERNETES
Chris Van Tuin Chief Technologist, NA West @chrisvantuin [email protected] -
DEV QA OPS THE AVERAGE ENTERPRISE DOES DEPLOYMENTS EVERY
6 TO 9 MONTHS. Walled off people, walled off processes, walled off technologies with surprisingly little to no automation -
HOW DOES I.T. ACCELERATE BUSINESS INNOVATION? 6 to 9
months Innovation Hours to Weeks -
ENABLING INNOVATION, WHILE EXECUTING AT SCALE Static & Planned Dynamic
& Policy Driven Execution Innovation Innovation Execution Old New -
+ ≠ TAXI TRANSPORTATION DISRUPTER https://goo.gl/MP7QQH Ack: Andrew Ng APP
IT’S NOT JUST A… + -
THE STRATEGIC DIFFERENTIATOR The Fab Powered by Automation “copy exactly”
The Software Factory Powered by Automation -
MOST THAN JUST SOFTWARE, THE DISRUPTERS… Empowered organization Speed
Up Innovation Time Change Move Fast, Break Things Culture of experimentation A 20% vs. 25% Shorten the Feedback Loop Real-time data-driven intelligence & personalization AI / ML Data, Data, Data B -
I.T. ENABLING BUSINESS DIFFERENTIATION Taxi Ridesharing $2.50/mile, $0.50/idle Fare
Static Pricing Dynamic Pricing vs. Ack: William Benton Fare based on real-time and historical data: supply/demand distance, time route, traffic etc. -
I.T. MUST TRANSFORM FROM A COST CENTER INTO AN
INNOVATION CENTER Powered by DevOps + Automation + + DEV QA OPS Culture Process Automation Technology Linux + Containers IaaS Orchestration CI/CD Source Control Management Collaboration Build and Artifact Management Testing Frameworks Cloud Native Applications Hybrid Cloud Open Source Agile, Iterative, Continuous, Infrastructure as Code Collaborative Transparent Open THE SOFTWARE FACTORY -
CONTAINERS Software packaging concept that typically includes an application and
all of its runtime dependencies • HIGHER quality software releases • SHORTER test cycles • EASIER application management CONTAINER CONTAINER APP LIBS HOST OS SERVER APP LIBS BENEFITS -
LAPTOP Container Application OS dependencies Guest VM LINUX BARE METAL
Container Application OS dependencies LINUX VIRTUALIZATION Container Application OS dependencies Virtual Machine LINUX PRIVATE CLOUD Container Application OS dependencies Virtual Machine LINUX PUBLIC CLOUD Container Application OS dependencies Virtual Machine LINUX CONTAINERIZED MICROSERVICES Build Once, Deploy Anywhere -
https://medium.com/@bibryam/cloud-native-container-design-principles-144ceaa98dba CLOUD NATIVE CONTAINER DESIGN PRINCIPLES: Build Time Does one
thing well Built with depenedencies, depends on Linux kernel Build once, deploy anywhere Single Concern Principle Self Containment Principle Image Imutability Principle -
Process Dispensability Principle CLOUD NATIVE CONTAINER DESIGN PRINCIPLES: Runtime Ephemeral
- short lived, replaceable Read & react to events APIs to obeserver & manage Resource requirements defined and restricted High Observability Principle Lifecycle Conformance Principle Runtime Confinement Principle -
docker.io Registry Private Registry FROM fedora:1.0 CMD echo “Hello” Build
file Physical, Virtual, Cloud Container Image Container Instance Build Run Ship CONTAINERS ENABLE DEVOPS CONTAINERS ENABLE DEVSECOPS FROM registry.redhat.com/rhel7 RUN groupadd -g 999 appuser && \ useradd -r -u 999 -g appuser appuser USER appuser CMD echo “Hello” -
CONTAINER IMAGE JAR CONTAINER IMAGE Application Application Language runtimes OS
dependencies 1.2/latest 1.1 -
ERGED SOFTWARE UPPLY CHAIN TAINER IMAGE CONTAINER IMAGE Application
Language runtimes OS dependencies 1.2/latest 1.1 TAINER IMAGE CONTAINER IMAGE Application Language runtimes OS dependencies 1.2/latest 1.1 TAINER IMAGE CONTAINER IMAGE Application Language runtimes OS dependencies 1.2/latest 1.1 SUPPLY CHAIN CONTAINER IMAGE JAR CONTAINER IMAGE Application Application Language runtimes OS dependencies 1.2/latest 1.1 CONVERGED SOFTWARE SUPPLY CHAIN Build file Container Image CONTAINER IMAGE JAR CONTAINER IMAGE Application Application Language runtimes OS dependencies Container Instance BARE METAL VIRTUAL PRIVATE CLOUD PUBLIC CLOUD Automated Software Factory Speed, Resiliency, Scalability, Security Configs / Storage Developer Apps / DB Operations -
BARE METAL VIRTUAL PRIVATE CLOUD PUBLIC CLOUD Automated Software Factory
Speed, Resiliency, Scalability, Security BARE METAL VIRTUAL PRIVATE CLOUD PUBLIC CLOUD Automated Software Factory Speed, Resiliency, Scalability, Security BARE METAL VIRTUAL PRIVATE CLOUD PUBLIC CLOUD Automated Software Factory Speed, Resiliency, Scalability, Security Speed, Agility, Resiliency, Scalability, Efficiency, Security -
DEMO APPLICATION Web App DEV QA OPS THE AVERAGE ENTERPRISE
DOES DEPLOYMENTS EVERY 6 TO 9 MONTHS. Walled off people, walled off processes, walled off technologies with surprisingly little to no automation DEV QA OPS THE AVERAGE ENTERPRISE DOES DEPLOYMENTS EVERY 6 TO 9 MONTHS. Walled off people, walled off processes, walled off technologies with surprisingly little to no automation -
Web Application replicas: 1, role: app image: myapp:1.0 replicas:
2, role: web image: httpd:1.7.9 ORCHESTRATION Speed & Agility: Declarative, Deployment Controller Manager & Data Store (etcd) -
Web Application ORCHESTRATION Declarative, Deployment Nodes Controller Manager & Data
Store (etcd) Physical, VM, Cloud Instances replicas: 2, role: web image: httpd:1.7.9 replicas: 1, role: app image: myapp:1.0 -
role: app role: web role: web Pods Nodes Image Registry
ORCHESTRATION Schedule + Provision Pods (Compute/Storage/Network) Web Application replicas: 2, role: web image: httpd:1.7.9 replicas: 1, role: app image: myapp:1.0 -
Web Application role: web role: app role: web replicas: 1,
role: app replicas: 2, role: web ORCHESTRATION Services (Load Balancer), Service discovery with selectors and pod labels Pods Nodes Services Controller Manager & Data Store (etcd) -
Web Application ORCHESTRATION Service (Load Balancer) Pods Nodes Controller Manager
& Data Store (etcd) Ingress / Routes role: web role: app role: web replicas: 1, role: app replicas: 2, role: web Services -
HEALTH CHECK Resiliency Monitoring & Logging Pods Nodes Services Web
Application role: web role: app role: web Ingress / Routes Health Check replicas: 1, role: app replicas: 2, role: web -
Pods Nodes Services Web Application role: web role: app role:
web replicas: 1, role: app replicas: 2, role: web role: web Controller Manager & Data Store (etcd) HEALTH CHECK Readiness Probe e.g. tcp, http, script Ingress / Routes -
Web Application replicas: 1, role: app replicas: 2,
role: web Pods Nodes Services role: web role: app role: web Controller Manager & Data Store (etcd) HEALTH CHECK Ingress / Routes -
Web Application Monitoring & Logging 80% CPU Pods Nodes Services
role: web role: app role: web Ingress / Routes replicas: 1, role: app replicas: 2, role: web AUTO-SCALE Scalability & Efficiency -
Web Application 80% CPU Pods Nodes Services role: web role:
app role: web Controller Manager & Data Store (etcd) role: app AUTO-SCALE Ingress / Routes replicas: 2 role: app replicas: 2, role: web -
Pods Nodes Services Web Application 50% CPU role: web role:
app role: app role: web Controller Manager & Data Store (etcd) AUTO-SCALE Ingress / Routes replicas: 2, role: web replicas: 2, role: app -
”only about 1/3 of ideas improve the metrics they
were designed to improve.” Ronny Kohavi, Microsoft (Amazon) MICROSERVICES RAPID INNNOVATION & EXPERIMENTATION -
25% Conversion Rate ?! Conversion Rate 100% Version B Version
A Ingress CANARY DEPLOYMENTS Tests / CI CANARY DEPLOYMENTS Build confidence in new version Service selector: app=demo version=A label: app=demo version=A 25% Conversion Rate ??% Conversion Rate -
25% Conversion Rate 30% Conversion Rate 75% 25% Version B
Version A Ingress CANARY DEPLOYMENTS CANARY DEPLOYMENTS Requires app to support side-by-side version Service Service selector: app=demo label: app=demo version=A 25% Conversion Rate % Conversion Rate label: app=demo version=B -
25% Conversion Rate 30% Conversion Rate 100% Version B Version
A Ingress CANARY DEPLOYMENTS Service selector: app=demo version=B label: app=demo version=A 25% Conversion Rate 30% Conversion Rate label: app=demo version=B -
EXTERNAL SERVICES Database outside cluster with IP address External Mongo
Database Service External Mongo Database Service Development Production IP=10.200.0.2 port=27017 IP=10.100.0.9 port=27017 -
EXTERNAL SERVICES Database outside cluster with IP address Pods Nodes
Services WebApp role=webapp replicas=2, role=webapp External Mongo Database Service IP=10.200.0.2 port=27017 Network External Mongo Database Service IP=10.100.0.9 port=27017 -
EXTERNAL SERVICES Database outside cluster with IP address Pods Nodes
Services WebApp role=webapp replicas=2, role=webapp External Mongo Database Service IP=10.200.0.2 port=27017 Network External Mongo Database Service IP=10.100.0.9 port=27017 Database name=mongo port=27017 targetport=27017 Endpoint IP=10.200.0.2 port=27017 Database kind=Service type=ClusterIP name=mongo port=27017 targetport=27017 -
EXTERNAL SERVICES Database outside cluster with IP address Pods Nodes
Services WebApp role=webapp replicas=2, role=webapp External Mongo Database Service IP=10.200.0.2 port=27017 Network External Mongo Database Service IP=10.100.0.9 port=27017 Database name=mongo port=27017 targetport=27017 Endpoint IP=10.200.0.2 port=27017 Connect with mongodb://mongo Database kind=Service type=ClusterIP name=mongo port=27017 targetport=27017 kind=Endpoints name=mongo ip=10.200.0.2 port=27017 -
EXTERNAL SERVICES Database outside cluster with IP address Pods Nodes
Services WebApp role=webapp replicas=2, role=webapp External Mongo Database Service IP=10.200.0.2 port=27017 Network External Mongo Database Service IP=10.100.0.9 port=27017 Database name=mongo port=27017 targetport=27017 Endpoint IP=10.100.0.9 port=27017 kind=Service type=ClusterIP name=mongo port=27017 targetport=27017 kind=Endpoints name=mongo ip=10.200.0.9 port=27017 Connect with mongodb://mongo Database -
Pods Nodes Services Database name: mongo type: ExternalName externalName: mongo52101.domain,.name
EXTERNAL SERVICES Using CNAME redirection mongodb:// <dbuser>: <dbpassword> @mongo:<port>/dev mongodb://<dbuser>:<dbpassword> @mongo52101.domain.name:52101/dev Cloud Mongo Database Service WebApp role=webapp replicas=2, role=webapp .name EXTERNAL SERVICE Connecting to Service with dynamic URI with a static ExternalName Kubernetes service -
PERSISTENT VOLUMES Host Container Host Container Host Container Data in
Container Data lost when Container terminates Data lost when Host terminates Independent of Container & Host Data in a Host Volume Networked Volume Data lost when Cloud instance (local) Data lost when Container terminates (image) Independent of Container & Cloud instance (EBS/network) DATA PERSISTENCE -
1. Maintains a sticky network ID/name across restarts e.g. mongo-0,
mongo-1, mongo-2 2. Ordered Operations with ordinal index e.g. name-0, name-1, name-2 3. Stable, persistent storage (linked to ordinal index/name) 4. Mandatory headless service (no single IP) for integrations KUBERNETES STATEFULSETS -
role=mongo type=leader Nodes Pods Services Mongo StatefulSet replicas=2 role=mongo Client
mongo-0 D A B C C DATABASE STATEFUL SETS StatefulSet with 2 replicas , headless service, direct access to pods pvc Read / Write Persistent Volume -
DATABASE STATEFUL SETS role=mongo type=leader role=mongo type=follower Nodes Pods Services
Client Mongo-0 Mongo-1 D A B C C Mongo StatefulSet replicas=2 role=mongo pvc pvc Read / Write Read / Only Persistent Volume -
role=mongo type=leader role=mongo type=follower role=mongo type=follower Nodes Pods Services Mongo-0
Mongo-1 Mongo-2 pvc pvc pvc Persistent Volume A B C C D Mongo StatefulSet replicas=3 role=mongo Read / Write Read / Only Read / Only DATABASE STATEFUL SETS Scale to 3 replicas Client -
role=mongo type=leader role=mongo type=follower role=mongo type=follower Nodes Pods Services Mongo-0
Mongo-1 Mongo-2 pvc pvc pvc Persistent Volume A B C D Mongo StatefulSet replicas=3 role=mongo DATABASE STATEFUL SETS Unresponsive Pod Client -
role=mongo type=leader role=mongo type=follower Nodes Pods Services Mongo-0 Mongo-1 pvc
pvc Persistent Volume A B D role=mongo type=follower Mongo-2 pvc C Mongo StatefulSet replicas=3 role=mongo DATABASE STATEFUL SETS Auto recovery Client -
Application v3 Development Application V2 Test Application v1 Production DB
v1 DB v2 DB v3 CI/CD PIPELINE Version control database updates, ex: flyway V3__add_table_scooter.sql V2__add_table_truck.sql V1__add_table_car.sql -
DATABASE MIGRATIONS Version control database updates with Containers CONTAINER IMAGE
CONTAINER BUILD FILE SQL MIGRATION SCRIPT Source Code Repository V2__add_table.sql Source Code Repository V2__add_table.sql /var/flyway/data Flyway flyway-mydb:v2.0.0 Registry + Dockerfile -
Nodes Pods Services postgresql-0 Persistent Volume A B D C
PostgreSQL StatefulSet replicas=1 role=postgresq pvcl DATABASE MIGRATION StatefulSet deployment with headless Service v1 -
Nodes Pods Services postgresql-0 Persistent Volume A B D C
PostgreSQL StatefulSet replicas=1 role=postgresql Pvc DATABASE MIGRATIONS Create a Job for Flyway Flyway Job Secrets = Database Connection Info v1 flyway-mydb:v2.0.0 Image Registry Flyway -
role=postgressql type=primary Nodes Pods Services postgresql-0 Persistent Volume A B
D C PostgreSQL StatefulSet replicas=1 role=postgresql pvc DATABASE MIGRATIONS Apply schema changes to database Flyway Job Secrets = Database Connection Info V2 flyway-mydb:v2.0.0 Flyway -
role=postgresql type=primary Nodes Pods Services postgresql-0 Persistent Volume A B
D C PostgreSQL StatefulSet replicas=1 role=postgresql Pvc DATABASE MIGRATIONS Version control for database with Kubernetes V2 -
MONITORING CONSIDERATIONS Kubernetes* Container* Host Cluster services, services, pods,
deployments metrics Container native metrics Traditional resource metrics - cpu, memory, network, storage prometheus + grafana kubernetes-state-metrics probes Stack Metrics Tool node-exporter Kubernetes metrics server: kubelet:cAdvisor Microservices Distributed applications - traditional app metrics - service discovery - distributed tracing prometheus + grafana jaeger tracing istio -
KUBERNETES NATIVE ADD-ONS knative github.com/knative istio istio.io Server- less Service
Mesh CI/CD tekton tekton.dev -
GENERAL DISTRIBUTION ISTIO BENEFITS Istio makes it easy to create
a network of deployed services with load balancing, service-to-service authentication, monitoring, and more, helping to avoid operational nightmares. POLICY ROUTING TELEMETRY Provides the observability needed to manage microservices, such as how services are invoked, communication flows, and points of latency Allows for the control of routing flows Grants the ability to write policy that applies to all applications and is not language specific -
GENERAL DISTRIBUTION INCREASING THE VALUE OF ISTIO VISUALIZATION TRACING Kiali
includes Jaeger Tracing, which provides distributed tracing out of the box. Kiali works with Istio to visualize the service mesh topology, features like circuit breakers or request rates. -
Traffic Control Service Resiliency Chaos Testing Observ- ability Security SERVICE
MESH WITH ISTIO Dedicated infrastructure layer for making service-to-service communication safe, fast, and reliable Deploy as a lightweight side-car network proxy -
CONFIDENTIAL - FOR INTERNAL USE ONLY MICROSERVICES WITHOUT ISTIO Container
JVM service A discovery load-balancer resiliency metrics tracing app logic JVM service B discovery load-balancer resiliency metrics tracing app logic Container JVM service C discovery load-balancer resiliency metrics tracing app logic -
CONFIDENTIAL - FOR INTERNAL USE ONLY MICROSERVICES WITH ISTIO Container
JVM service C app logic Pod Sidecar Container Envoy Container JVM service A app logic Pod Sidecar Container Envoy Container JVM service B app logic Pod Sidecar Container Envoy -
ISTIO SERVICE MESH Envoy istio-ingress Envoy App A Envoy App
B Envoy App C istio-pilot istio-mixer istio-auth HTTP Req/Resp Kubernetes Pods Istio Components Config to Envoy Access Control and Telemetry -
24% 76% v1 v2 apiVersion: config.istio.io/v1alpha2 kind: RouteRule metadata: name:
recommendation-v1-v2 spec: destination: namespace: tutorial name: recommendation precedence: 5 route: - labels: version: v1 weight: 76 - labels: version: v2 weight: 24 CANARY RELEASE BY WEIGHT RouteRule #2: Route 94% to v1 and 6% to v2 -
“.*Safari.*” Default v1 v2 apiVersion: config.istio.io/v1alpha2 kind: RouteRule metadata: name:
recommendation-safari spec: destination: namespace: tutorial name: recommendation precedence: 2 match: request: headers: user-agent: regex: ".*Safari.*" route: - labels: version: v2 ROUTING BY HEADER By Geography, Mobile Device, Browser, Customer, … RouteRule #3: Route “Safari” to v2 -
v1 v2 apiVersion: config.istio.io/v1alpha2 kind: RouteRule metadata: name: recommendation-mirror spec:
destination: namespace: tutorial name: recommendation precedence: 2 route: - labels: version: v1 weight: 100 - labels: version: v2 weight: 0 mirror: namespace: tutorial name: recommendation labels: version: v2 DARK LAUNCH Mirror production traffic for pre-release testing RouteRule #3: Route “Safari” to v2 Mirror Production Traffic To v2 100% 100% Test Production -
apiVersion: config.istio.io/v1alpha2 kind: EgressRule metadata: name: httpbin-egress-rule spec: destination: service:
httpbin.org ports: - port: 80 protocol: http SECURE BY DEFAULT Egress blocks all traffic unless unless whitelisted with EgressRule EgressRule: Allow httpbin.org:80 (http) role=web Pods Nodes http://httpbin.org Istio EgressRule