Tweet
Tweet

Slide 1

Slide 1 text

HOW CAN YOU ACHIEVE WEB-SCALE DATA CENTER AUTOMATION? Michael Hausenblas, Mesosphere Inc

Slide 2

Slide 2 text

© 2016 Mesosphere, Inc. All Rights Reserved. HOW CAN YOU ACHIEVE WEB- SCALE DATA CENTER AUTOMATION? 2 Michael Hausenblas, Developer & Cloud Advocate | 2016-03-15 | CeBIT

Slide 3

Slide 3 text

© 2016 Mesosphere, Inc. All Rights Reserved. SDDC CHALLENGES 3

Slide 4

Slide 4 text

© 2016 Mesosphere, Inc. All Rights Reserved. SDDC CHALLENGES 4

Slide 5

Slide 5 text

© 2016 Mesosphere, Inc. All Rights Reserved. SDDC CHALLENGES 5

Slide 6

Slide 6 text

© 2016 Mesosphere, Inc. All Rights Reserved. SDDC CHALLENGES 6

Slide 7

Slide 7 text

© 2016 Mesosphere, Inc. All Rights Reserved. SDDC CHALLENGES 7 • Software is eating the world, need to release features faster (and more of them) than your competitor • Mobile devices, constraint devices, IoT • Distributed applications

Slide 8

Slide 8 text

© 2016 Mesosphere, Inc. All Rights Reserved. IT'S TIME FOR A NEW SORT OF OPERATING SYSTEM 8

Slide 9

Slide 9 text

© 2016 Mesosphere, Inc. All Rights Reserved. AN OPERATING SYSTEM … 9 … provides abstraction of resources … exposes common services … enables faster app development

Slide 10

Slide 10 text

© 2016 Mesosphere, Inc. All Rights Reserved. RESOURCE ABSTRACTION 10 counter = 1

Slide 11

Slide 11 text

© 2016 Mesosphere, Inc. All Rights Reserved. COMMON SERVICES 11 • memory management • job scheduling • filesystem • networking stack • access control • user management • package manager

Slide 12

Slide 12 text

© 2016 Mesosphere, Inc. All Rights Reserved. COMMON SERVICES 12 int socket(int domain, int type, int protocol); int bind(int sockfd, struct sockaddr *my_addr, int addrlen); int connect(int sockfd, struct sockaddr *serv_addr, int addrlen); int listen(int sockfd, int backlog); ping netstat ethtool traceroute ifconfig

Slide 13

Slide 13 text

© 2016 Mesosphere, Inc. All Rights Reserved. MEET THE DATACENTER
 OPERATING
 SYSTEM
 13

Slide 14

Slide 14 text

© 2016 Mesosphere, Inc. All Rights Reserved. A LITTLE HISTORY 14 PHYSICAL (x86) VIRTUAL HYPERSCALE MAINFRAME SERVER VIRTUAL MACHINE PARTITION (LPAR) UNIT OF INTERACTION " ERP, CRM, PRODUCTIVITY, MAIL & WEB SERVER " LINUX, WINDOWS " DATA / TRANSACTION PROCESSING " UNIX, IBM OS/360 DEFINITIVE APPS AND OS " ERP, CRM, PRODUCTIVITY, MAIL & WEB SERVER " HYPERVISOR + GUEST OS " BIG DATA, INTERNET OF THINGS, MOBILE APPS " ??? ??? DATACENTER " BIG DATA, INTERNET OF THINGS, MOBILE APPS " DATACENTER OPERATING SYSTEM (DCOS)

Slide 15

Slide 15 text

© 2016 Mesosphere, Inc. All Rights Reserved. TRADITIONAL VS. DCOS APPROACH 15 • Many “snowflakes” • Management nightmare • Lengthy cycles to deploy code • Low utilization Mesosphere DCOS DCOS APPROACH Container
 Apps
 (All) Big Data Analytics 
 (All) PaaS 
 (All) Stateful Service
 (All) • High performance and resource isolation • Easy scalability and multi-tenancy • Fault tolerant and highly available • Highly efficient with highest utilization proven in mission- critical deployments on-prem or in the cloud TRADITIONAL APPROACH PaaS 
 1 Container
 App 
 1 Big Data Analytics 1 Big Data Analytics 2 PaaS 
 2 Container
 App 
 2 Stateful Service
 1 Stateful Service
 2 workload portability

Slide 16

Slide 16 text

© 2016 Mesosphere, Inc. All Rights Reserved. 16

Slide 17

Slide 17 text

© 2016 Mesosphere, Inc. All Rights Reserved. BENEFITS 17 DCOS • Run stateless services (Web server, app server, etc.) and Big Data services like Kafka, Spark, or Cassandra together on one cluster • Dynamic partitioning of your cluster, depending on your business requirements • Increased utilization (10% → 80%++)

Slide 18

Slide 18 text

© 2016 Mesosphere, Inc. All Rights Reserved. CONTAINER OPERATIONS 18

Slide 19

Slide 19 text

© 2016 Mesosphere, Inc. All Rights Reserved. CONTAINER OPERATIONS 19 • CI/CD (delivery, versioning, smoke tests) • container orchestration (failover, support for multiple workload types, stateful services) • troubleshooting (monitoring/logging/debugging/ tracing) • high utilization • security (isolation, authz/authn)

Slide 20

Slide 20 text

© 2016 Mesosphere, Inc. All Rights Reserved. CONTAINER LIFE CYCLE 20

Slide 21

Slide 21 text

© 2016 Mesosphere, Inc. All Rights Reserved. CONTAINER ORCHESTRATION 21

Slide 22

Slide 22 text

© 2016 Mesosphere, Inc. All Rights Reserved. WHAT DOES A SCHEDULER DO? 22

Slide 23

Slide 23 text

© 2016 Mesosphere, Inc. All Rights Reserved. DEMO 23 https://www.youtube.com/watch?v=u4fEeIrLJZ4

Slide 24

Slide 24 text

© 2016 Mesosphere, Inc. All Rights Reserved. TAKE- AWAYS 24 • To tackle SDDC challenges, a new approach is necessary: a distributed operating system • Portable workloads across environments
 (on-premises/hybrid/public cloud) are possible • Container operations is becoming mainstream

Slide 25

Slide 25 text

© 2016 Mesosphere, Inc. All Rights Reserved. Q & A 25 • @mhausenblas • mhausenblas.info • @mesosphere • mesosphere.com