FreeBSD Operations at Limelight Networks (part 2 of 2)

Video in TIB AV-Portal: FreeBSD Operations at Limelight Networks (part 2 of 2)

Formal Metadata

Title
FreeBSD Operations at Limelight Networks (part 2 of 2)
Subtitle
An Overview of Operating at Internet Scale
Title of Series
Author
License
CC Attribution - ShareAlike 3.0 Unported:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal and non-commercial purpose as long as the work is attributed to the author in the manner specified by the author or licensor and the work or content is shared also in adapted form only under the conditions of this license.
Identifiers
Publisher
Release Date
2015
Language
English

Content Metadata

Subject Area
Abstract
In this talk, we'll look at Limelight's global CDN architecture and the practice of large scale web operations with FreeBSD. We'll investigate how FreeBSD makes these tasks easier and the strategies and tools we've developed to run our operations. We'll then look at why the engineering team chose SaltStack to further improve our operations capabilities and reduce deployment and fault handling times. Finally, we'll finish up with an overview of metrics and monitoring at scale with Zabbix and OpenTSDB. Limelight Networks is one of the "Big Three" CDNs and runs its edge using FreeBSD.
laptops effective state time machines jump drop Heads part central Mathematical single touch box partitions Systems form feedback moment instances lines Application configuration management maintainability Cache types job Computer animation environmental networks Case hard drives configurable route
and wrote punitive if we don't some of all so that the biggest thing running on DSP is ah caching software which is a C application and then there's a bunch of maintenance software on a lot of its Pearl other that does things like the shipping logs off to be processed and another configuration changes and we do have some other stuff in the works of trying to think most of its see action on our obvious the edge we don't we don't have anything else going on at the moment of truth ruins on that so a very deep Christian I think so we've got you know that that the 1st thing is we wanna keep our boxes patched and is the BEST team we do a better job of this and anybody else because we got this release process and we've got the put their stuff on further support stuff on our older systems right now we can order our systems for forms and and make sure that the patched up a lot of the monitoring stuff is you know by syslog so we're getting those log shipped off of the machine so they can be tampered with other central aggregation hosts and then but a lot of its probably human right now but we do have tripwires and stuff in place with configuration management and and with other systems as far as deployment tour light on yes OK so that's actually why we have this big feedback loop we wanna make that not such a big deal and so will will look for instance Paul the MFS or whatever with live on fixed to ll stable branch will but will do a build and then will re-image all of our machines we don't we don't use previously anything like that it's all image-based for for this stuff yeah I didn't touch on it in the presentation but we're using UFS on these machines because we're doing valence place and if you've ever seen 1 of the Netflix talks are boxes look just like that you know we've got hard drives with single partitions narrator anything when 1 of those drops we just leave it dead for a long time i z of us would be interesting for some other use cases where we you know if we start to moving some of our other applications on BST on and you know I've used food environments I think there are some say that saved me on my laptop from broken updates that does the same thing in in those types of roles or yes so we're all of these edge boxes are running dual-stack but I don't know a lot about the Jason run outta here G are we using their what's in rout how did they get their IPV 6 united States states per machine it's actually pretty small I ratio for us today but we do supported by the by I think we use our syslog on previously in those are just going to aggregators in each of our part so I and I don't know what actually what happens after that I'm not evolve with those machines it's just a line and are configured that we get offer them yet so I don't know off the top of my head anything else could make use
Feedback