lessons in moving from physical hosts to mesos
TRANSCRIPT
![Page 1: Lessons in moving from physical hosts to mesos](https://reader035.vdocuments.site/reader035/viewer/2022062902/58ef6e161a28ab061b8b45a1/html5/thumbnails/1.jpg)
Lessons in moving from physical hosts to Mesos
Raj Shekhar, Senior Site Reliability Engineer
@ilunatech
![Page 2: Lessons in moving from physical hosts to mesos](https://reader035.vdocuments.site/reader035/viewer/2022062902/58ef6e161a28ab061b8b45a1/html5/thumbnails/2.jpg)
Mesos
WHATWHYHOW
NOW WHAT
![Page 3: Lessons in moving from physical hosts to mesos](https://reader035.vdocuments.site/reader035/viewer/2022062902/58ef6e161a28ab061b8b45a1/html5/thumbnails/3.jpg)
How most Ops teams run clusters today
![Page 4: Lessons in moving from physical hosts to mesos](https://reader035.vdocuments.site/reader035/viewer/2022062902/58ef6e161a28ab061b8b45a1/html5/thumbnails/4.jpg)
Static partitioning has problemsUnequal load distribution on machinesSlower to add capacityNot fault tolerant
![Page 5: Lessons in moving from physical hosts to mesos](https://reader035.vdocuments.site/reader035/viewer/2022062902/58ef6e161a28ab061b8b45a1/html5/thumbnails/5.jpg)
Is there a better way?Do we want machines or do we want resources?
![Page 6: Lessons in moving from physical hosts to mesos](https://reader035.vdocuments.site/reader035/viewer/2022062902/58ef6e161a28ab061b8b45a1/html5/thumbnails/6.jpg)
MesosResource manager - the datacenter is one big poolCan run multi-tenant workloadsFailure detectionServices are isolated from one another
![Page 7: Lessons in moving from physical hosts to mesos](https://reader035.vdocuments.site/reader035/viewer/2022062902/58ef6e161a28ab061b8b45a1/html5/thumbnails/7.jpg)
Why Mesos - Better resource utilizationRun multi-tenant workload on machines
Dynamic partitioning - no dedicated machines for tasks
Less resource hungry than virtual machines
![Page 8: Lessons in moving from physical hosts to mesos](https://reader035.vdocuments.site/reader035/viewer/2022062902/58ef6e161a28ab061b8b45a1/html5/thumbnails/8.jpg)
Why Mesos - all the other good things
Fault tolerant - automatically restart failed jobs
Elasticity - grow and shrink on demand
Faster deploys
![Page 9: Lessons in moving from physical hosts to mesos](https://reader035.vdocuments.site/reader035/viewer/2022062902/58ef6e161a28ab061b8b45a1/html5/thumbnails/9.jpg)
T.co - URL shortening
http://example.com/example http://t.co/examp
![Page 10: Lessons in moving from physical hosts to mesos](https://reader035.vdocuments.site/reader035/viewer/2022062902/58ef6e161a28ab061b8b45a1/html5/thumbnails/10.jpg)
How
Package Deploy Test Go Live!
![Page 11: Lessons in moving from physical hosts to mesos](https://reader035.vdocuments.site/reader035/viewer/2022062902/58ef6e161a28ab061b8b45a1/html5/thumbnails/11.jpg)
Life after Go LiveLowered operating expenseFewer routine operational tasksFaster deploys
![Page 12: Lessons in moving from physical hosts to mesos](https://reader035.vdocuments.site/reader035/viewer/2022062902/58ef6e161a28ab061b8b45a1/html5/thumbnails/12.jpg)
Job throttling
Sudden spikes in latencies
What we learned
cgroups and cpu quotas
![Page 13: Lessons in moving from physical hosts to mesos](https://reader035.vdocuments.site/reader035/viewer/2022062902/58ef6e161a28ab061b8b45a1/html5/thumbnails/13.jpg)
Capacity planning
Max traffic of the cluster was lower than our expectationWhat we learned
Different CPU variants have different throughput
![Page 14: Lessons in moving from physical hosts to mesos](https://reader035.vdocuments.site/reader035/viewer/2022062902/58ef6e161a28ab061b8b45a1/html5/thumbnails/14.jpg)
Rethink service discovery
Services get hosts and ports assigned dynamically
What we learned
Use static proxies to forward connections
![Page 15: Lessons in moving from physical hosts to mesos](https://reader035.vdocuments.site/reader035/viewer/2022062902/58ef6e161a28ab061b8b45a1/html5/thumbnails/15.jpg)
No perfect isolation
Sudden spike in latency
What we learned
Async ops where possible, noisy neighbours still affect us