Building self-service on demand infrastructure with Puppet and VMware
Apr. 1, 2013•0 likes
3 likes
Be the first to like this
Show More
•6,660 views
views
Total views
0
On Slideshare
0
From embeds
0
Number of embeds
0
Download to read offline
Report
Technology
"Building self-service infrastructure with Puppet and VMware: An intro to vSphere automation" by Cody Herriges of Puppet Labs at Puppet Camp London 2013. Find a Puppet Camp near you: https://puppetlabs.com/community/puppet-camp/
Really about…
• A little about Operations at Puppet Labs.
• Why we choose on premise virtualization over
giving everyone EC2/Rackspace credentials.
• How we’re tackling the ramifications of that ^
decision.
2 | CONFIDENTIAL & PROPRIETARY
Who am I?
• Cody Herriges
• Operations Engineer
• Joined Puppet Labs in 2010
• Been using Puppet for 5.5 years
• Currently responsible for internal Virtualization
and Storage architectures.
4 | CONFIDENTIAL & PROPRIETARY
Puppet Labs Operations
• Three senior infrastructure engineers.
• Two desktop support engineers.
• Two Jr-ish infrastructure engineers.
• Both start while I am here in Europe.
5 | CONFIDENTIAL & PROPRIETARY
Our current strategy
• In the cloud?
• Our production web applications; forge,
redmine, ask, puppetlabs.com.
• All other production?
• On premise data center.
6 | CONFIDENTIAL & PROPRIETARY
Defining focus
• We were always chasing things that were
“broken”.
• We weren’t making any personal growth.
• No time to innovate.
• Infrastructure design and implementation never
matured.
7 | CONFIDENTIAL & PROPRIETARY
VMware
• Our implementation was holding us back.
• Rebuilt everything.
• Networking infrastructure provided by 1GbE Cisco
and 10GbE Juniper.
• Storage over iSCSI, backed by EMC VNX and
Nexenta.
8 | CONFIDENTIAL & PROPRIETARY
IaaS
• Pros
• Operational over a capital investment.
• You get off the ground fast.
• Don’t have to worry about power, rack space,
cooling, cable management, networking.
• On demand unlimited resources at your finger
tips.
9 | CONFIDENTIAL & PROPRIETARY
IaaS
• Cons
• Performance is expensive.
• Scale is expensive.
• Redundancy can increase cost and/or
application complexity.
• You lose most of your control over networking.
10 | CONFIDENTIAL & PROPRIETARY
Internal Infrastructure
• Pros
• Performance and scale per dollar is much less.
• Redundancy is loads simpler when you control
layer 2.
• Infrastructure can fit into the way you work
and develop.
• During an outage there is always something
you can do about it.
11 | CONFIDENTIAL & PROPRIETARY
Internal Infrastructure
• Cons
• You have to buy things.
• There is a datacenter to manage.
• Initial roll out will take time.
12 | CONFIDENTIAL & PROPRIETARY
Internal Infrastructure
• Other considerations
• If you have a physical office you’ll will have
some amount of internal infrastructure.
• There will always be a point where you out
grow that converted coat closet.
• The product you develop may require it.
13 | CONFIDENTIAL & PROPRIETARY
Loading ESXi
• vSphere hypervisor (ESXi) uses kickstart.
• Razor, capable of loading ESXi to non-volatile
storage.
• vCenter’s built in gPXE system, vSphere Auto
Deploy.
• We’ve used both.
14 | CONFIDENTIAL & PROPRIETARY