Mesosphere Datacenter Operating System Deploying Applications on DC/OS Keith McClellan - Technical Lead, Federal Programs keith.mcclellan@mesosphere.com V6
THE FUTURE IS ALREADY HERE IT S JUST NOT EVENLY DISTRIBUTED - WILLIAM FORD GIBSON 2
Datacenter Operating System (DC/OS) HYPERSCALE COMPUTING FOR MAINSTREAM ENTERPRISES NEW FORM FACTOR FOR DEVELOPING AND RUNNING APPS Unit of Management Partition (LPAR) Physical Server Logical Server (VM) Logical Datacenter MAINFRAME PHYSICAL (x86) VIRTUAL HYPERSCALE Definitive Apps Data Processing ERP, CRM Web Client / Server Modern Distributed Apps: Big Data, Web Scale, IoT Enabling Technology RISC Processor X86 HW & OS Hypervisor (Partitioning) Datacenter Operating System (Aggregation) 3
DC/OS KERNEL APPLIES LESSONS FROM EARLY INNOVATORS Production-proven Web Scale Cluster Managers Borg/Omega Tupperware/Bistro Apache Mesos and DC/OS ~2001 ~2007 2009+ Proprietary Proprietary Open Source (Apache License) Built at UC Berkeley AMPLab by Ben Hindman (Mesosphere Co-founder) Built in collaboration with Google to overcome some Borg Challenges Production proven at scale +80K hosts @ Twitter
HYPERSCALE MEANS: MODERN APPLICATIONS Functions & Logic Big Data + Analytics Engines Streaming Batch Microservices (in containers) Analytics Machine Learning Search Time Series Databases SQL / NoSQL
ANATOMY OF A MODERN APP Functions & Logic Big Data + Analytics Engines X Big Data Processing Data Storage Message Queue X X X Anything else
Hyperscale Computing Properties CONTAINER ORCHESTRATION Running Multiple Microservices-based Apps REST APIs Dependencies Between Microservices
IMPLICATIONS OF HYPERSCALE VIRTUAL HYPERSCALE Workload pooling Partition physical servers to run multiple VMs Aggregate a logical DC to run multiple distributed systems Deploying apps Place some VMs in server pool (e.g., DB, App Svr, Web Svr) Launch hundreds of microservices; service discovery automation Ensuring performance Increase VM size; clear host of other VMs Scale distributed microservices programmatically Business continuity Restart VMs in right order Ensure microservices are fault tolerant and self-healing
MODERN APPS RUNNING ON DC/OS Datacenter Operating System (DC/OS) App Functions Big Data + Analytics Engines Big Data Processing Data Storage Two-level scheduling Message Queue Container Orchestration Stateless and Stateful Services Operations & Service Quality Automation Reactive, Self-Healing Infrastructure Abstraction & Aggregation Servers VMs Cloud (Azure / Amazon / GCE) Turnkey setup & operations Single logical datacenter
MESOSPHERE DC/OS: HYPERSCALE COMPUTING, ENTERPRISE AGILITY The Modern Cloud Stack Modern Apps (Big Data built-in) Analytics & Database Streaming, Pipelines, Search, ML Relational, NoSQL, Data Warehouses Continuous Integration & Continuous Delivery Containers & Microservices Orchestration Mesosphere Infinity (Spark, Kafka, Cassandra, Akka) Common CI/CD Tools on Mesosphere DC/OS (GitHub, Jenkins) Mesosphere DC/OS (Mesos, Marathon, Datacenter Operations) DATA AGILITY DEVELOPER AGILITY HYPERSCALE OPERATIONS Hybrid Infrastructure (Compute, Storage, Networking) 10
Strong Capability DC/OS EXTENSIBILITY FAR BEYOND COMPETITION Some Capability Not Supported Mesosphere DC/OS Docker DUCP CoreOS Tectonic Redhat OpenShift Cloud Foundry Mesos Swarm (Production proven) (Under development) Kubernetes OpenShift Cloud Foundry (Nascent) (Production proven) (Production proven) Next Gen Application Support Distributed Services (e.g., Kafka, Cassandra) Containers (e.g., Docker, rkt) PaaS Big Data (e.g., Hadoop, Spark, Storm) Extensible Architecture Core Technology Kubernetes (Nascent) Supported Infrastructure On Premise Public Cloud
RECENT PARTNERSHIP ANNOUNCEMENTS 12
DEMO TIME!!! Get DC/OS at dcos.io (Apache License) Demo code available at http://github. com/mesosphere/iot-demo 13