scale of Kubernetes Not Part of Kubernetes Kubernetes solves the application orchestration problem Container Granular Volume Provisioning and High Availability Backup and Recovery Business Continuity and Disaster Recovery Data Encryption and Security Automated Storage Monitoring and Management
2019 PORTWORX | CONFIDENTIAL: DO NOT DISTRIBUTE PX-CENTRAL PX-CENTRAL PORTWORK RUNS ON ANY STORAGE HARDWARE PORTWORK RUNS ON ANY KUBERNETES PLATFORM PX-CENTRAL PX-MIGRATE PX-DR PX-STORE PX-BACKUP PX-AUTOPILOT PX-SECURE Data Platform for Kubernetes
back-office apps responsible for T-Mobile store experience during Apple launches SOLUTION ► Portworx Enterprise running on-prem to provide complete data lifecycle management- provisioning, encryption, backup, and more. RESULTS ► Regularly withstand the launch of new Apple products results in dramatic spike in usage Data services used:
Zero downtime for Kubernetes apps even in face of entire data center loss SOLUTION ► Two Kubernetes clusters- Production and DR site ► PX-DR provides cross data center replication of data and app config RESULTS ► Zero RPO DR with < 2 minute failover in event of primary data center loss Data services used:
Workloads ◦ On-Prem, not just for the cloud: K8s is transforming the data center on-prem: Ford, T-Mobile, Comcast, Lloyds Bank, Bank of America, ◦ Stateful Apps: Stateful apps are as easy as deploying stateless apps with Kubernetes extensions for storage and data. Stateful apps using databases (Postgres, Cassandra, Mongo, Redis etc), Elasticsearch, Kafka, Streaming data with Spark are deployed seamlessly with CSI and storage overlays ◦ App lifecycle Management: the lifecycle of containerized apps will be managed by K8s with HA, DR, Backup and Compliance extensions of K8s. • Tomorrow: Kubernetes will manage Apps and Infrastructure for ALL workloads ◦ New Control Plane for the Datacenter and Cloud: k8s is on its way to managing apps and infrastructure. CSI, CNI, Kubervirt extensions allow K8s to manage, storage networking and VMs. VMWare’s Project Pacific is an example. ◦ Automation Enabler for AI/ML: K8s will automate both apps and infrastructure enabling ML/AI mining of data and intelligent real-time services.
Provide a PaaS experience: GUI-based, services catalog, RBAC, built-in security, rapid on-ramping of new apps, scaling to production, manage app lifecycle, App-Granular Control: manage infrastructure and apps via Kubernetes with container granular controls vs machine centric controls Pay-for-Use: subscription model, monitoring for chargebacks Multi-cloud and heterogeneous: runs everywhere, agnostic to different hardware for compute, storage and networking, Design for Devops, Scale for IT Ops: build for a automated software defined devops experience but provide scale, security and reliability for IT Ops. Datalifecycle and App Insights: manage HA, DR, Backup, and Compliance across the app lifecycle. Capture app insights to ease deployment, drive optimization, and apply ML/AI tools. Enable higher Business Value: automation for agility and cost reduction; enable ML/IT to mine data for competitive advantage
pressures are driving Cloud Native initiatives DATA EXPLOSION 6 BILLION connected individuals by 2025 Source: IDC Data Age 2025 Source: ITRC Breach Statistics 2006-2016 SECURITY THREAT 600% increase in annual data breaches in last decade Source:Puppet State of DevOps SPEED IMPERATIVE 25X faster recovery by high-performing DevOps teams . YOUR BUSINESS TRENDS
are how Cloud native happens 95% of new apps use containers - via DevOps Source: 451research.com CLOUD NATIVE IS THE PLATFORM ADOPTION OF CLOUD-NATIVE TECHNOLOGIES CONTAINERS n=464 KUBERNETES n=477 78% deployed within two years 69% deployed within two years TRENDS
What Started This? ENTERPRISES ARE: APP 1 APP 2 APP 3 APP 4 VM-Cloud Public Cloud ON-PREM 1 Building a Container Native Platform • Development Agility • Automation • Low touch Ops 2 Seeking to run Applications Anywhere 3 Accelerating their Business through Agile Devops JOURNEY
AGILITY You are only as agile as the least nimble element in your platform. Kubernetes depends on agile infrastructure services to ensure its full functionality.
In fact, it defines the game through a new paradigm Infrastructure Services [Storage | Network | Security | DR] • YAML defined workflows via kubectl • CRDs • Declarative Deployments • GITOPS KUBERNETES EVOLUTION • A new lexicon for managing the app life cycle • Declarative approach driven through code and automation • Extensible Automation through CRDs and Operators
APP DEFINED CONTROL PLANE Machine Defined vs Application Defined MACHINE DEFINED CONTROL PLANE KUBERNETES EVOLUTION HOST … HOST KUBERNETES Storage ► Focus is on application aware infrastructure provisioning and lifecycle management ► Driven by an end user (application owner) Storage Storage Storage … Storage Storage Storage x86 Servers UBUNTU RHEL CENTOS LUNS VMWARE ► Focus is on machine provisioning and lifecycle management ► Driven by a VM, Storage and Network Admin
Overlay Built for Containers BUILDING A CONTAINER NATIVE PLATFORM PX PX PX Node 1 Node 2 Node 3 Virtualize Physical Storage and form a Container Storage Overlay Dynamic Container Granular Volume Provisioning driven through Kubernetes Container Virtual Volume Highly Available Anywhere
Scalable, Secure and Extends to Anywhere BUILDING A CONTAINER NATIVE PLATFORM PX PX Data Center 1 Data Center 2 Key Store High Density - Thousands of Application PODs per cluster Extends to Any Cloud PX Cloud Drive End to End Data Security Driven through Kubernetes
Building a Hybrid Cloud PaaS x86 x86 On Prem Data Center IBM Public Cloud Cloud Other Public Cloud On Prem OpenShift EKS OpenShift in Other Cloud Best in Class Integrated App and Data Experience USE CASES Top Use Case
Business Continuity and DR Cross Cloud Incremental Backup & Recovery of Applications and Data Hong Kong Any Cloud Asynchronous Data Replication (snapshots) Kubernetes Cluster 3 x86 x86 x86 x86 Data Center 1 Data Center 2 Kubernetes Cluster 2 Kubernetes Cluster 1 Synchronous Data Replication Metro Area 1 x86 x86 x86 x86 Metro Area 2 REPLICATION TYPE RPO/RTO ▶ Incremental Backup/Restorer ▶ RPO 15 minutes, fast RTO REPLICATION TYPE RPO/RTO ▶ Synchronous ▶ RPO Zero, fast RTO USE CASES
Multi Kubernetes ML/AI Workflow USE CASES ON PREM KUBERNETES CLUSTER GOOGLE CLOUD GPU ENABLED KUBERNETES GKE Worker Nodes Worker Nodes Master Config Airflow Config Spark Job Results SPARK Master SPARK Master Worker Nodes Worker Nodes Transient Source Data Results Data 1. Two OpenShift clusters on a single stretched portworx cluster 2. Airflow runs only on OPENSHIFT-RENO, In RENO namespace, in case of DR, Airflow in OPENSHIFT-TAHOE requires activation. 3. Use pre-provisioned volumes to store results on shared volume by both Openshift Clusters ON-PREM
So where is all this headed? Kubernetes Defined Data Centers Infrastructure Services [Storage | Network | Security | DR] A single Unified Control Plane for VMs & Containers VMs, EC2 managed directly by Kubernetes Run containers directly on k8s bare metal Container & VM Registry Application Data Backup and DR CONCLUSION
YOU PX Docs and Install http://docs.portworx.com http://install.portworx.com PX Asynchronous DR https://drive.google.com/open?id=1fnQM32b0CrcRrp_4cyTI9nxm2ONIXGTK PX-Metro (Synchronous DR) https://drive.google.com/open?id=1q90HWaZ9swkMqBswUBI0Y8u4RjoMFZ6i