Deploy in the cloud

Pablo Cantero @pablocantero Deploy in the Cloud

We were happy with our deploy process

Featured branches, Pull Requests, Continuous Integration etc, all that stuff.

Edson (Hudson)

Deploy made by push master deploy into the staging production
deploy into the production

$ git pull $ git checkout production $ git rebase
master $ git push origin production

But... where is the CLOUD?

Would the Cloud must be sexier?

•Launch Configuration •Auto Scaling Group •Scaling Policies •Metrics •Alarms Auto
Scaling

WARNING... As a programer the next slides with commands would
appear beautiful, but don’t trust them, they can be very dangerous

as-‐create-‐launch-‐config my_autoscale_config -‐-‐image-‐id ami-‐XXXXXXXX -‐-‐instance-‐type m1.small
-‐-‐group "My Security Group Name"

as-‐create-‐auto-‐scaling-‐group my_autoscale_group -‐-‐availability-‐zones us-‐east-‐1X -‐-‐launch-‐configuration my_autoscale_config
-‐-‐min-‐size 1 -‐-‐max-‐size 3 -‐-‐load-‐balancers my_load_balancer_name -‐-‐health-‐check-‐type ELB -‐-‐grace-‐period 300

as-‐put-‐scaling-‐policy ScaleUp -‐-‐auto-‐scaling-‐group my_autoscale_group -‐-‐adjustment=1
-‐-‐type ChangeInCapacity -‐-‐cooldown 300

mon-‐put-‐metric-‐alarm AlarmName -‐-‐comparison-‐operator value
-‐-‐evaluation-‐periods value -‐-‐metric-‐name value -‐-‐namespace value -‐-‐period value -‐-‐statistic value -‐-‐threshold value [-‐-‐actions-‐enabled value ] [-‐-‐alarm-‐actions value[,value...] ] [-‐-‐alarm-‐description value ] [-‐-‐dimensions "key1=value1,key2=value2..." ] [-‐-‐ok-‐actions value[,value...] ] [-‐-‐unit value ] [-‐-‐insufficient-‐data-‐actions value[,value...] ]

They can be very dangerous

require "aws-‐sdk" class QueueSizeMetric def initialize
AWS.config access_key_id: "…", secret_access_key: "…" end def put metric = AWS::CloudWatch::Metric.new "Worker", "QueueSize" size = SidekiqMetric.new.queue_size metric.put_data [{value: size}] end end

10 instances per 10 hours costs the same as 100
instances per 1 hour Rules in the Cloud

Talk is cheap, show me in practice...

Pablo Cantero @pablocantero Thank you

Deploy in the cloud

Deploy in the cloud

Pablo Cantero

More Decks by Pablo Cantero

Other Decks in Programming

Featured

Transcript

Pablo Cantero @pablocantero Deploy in the Cloud

We were happy with our deploy process

Featured branches, Pull Requests, Continuous Integration etc, all that stuff.

Edson (Hudson)

Deploy made by push master deploy into the staging production

$ git pull $ git checkout production $ git rebase

But... where is the CLOUD?

Would the Cloud must be sexier?

•Launch Configuration •Auto Scaling Group •Scaling Policies •Metrics •Alarms Auto

WARNING... As a programer the next slides with commands would

as-‐create-‐launch-‐config my_autoscale_config -‐-‐image-‐id ami-‐XXXXXXXX -‐-‐instance-‐type m1.small

as-‐create-‐auto-‐scaling-‐group my_autoscale_group -‐-‐availability-‐zones us-‐east-‐1X -‐-‐launch-‐configuration my_autoscale_config

as-‐put-‐scaling-‐policy ScaleUp -‐-‐auto-‐scaling-‐group my_autoscale_group -‐-‐adjustment=1

mon-‐put-‐metric-‐alarm AlarmName -‐-‐comparison-‐operator value

They can be very dangerous

require "aws-‐sdk" class QueueSizeMetric def initialize

10 instances per 10 hours costs the same as 100

Talk is cheap, show me in practice...

Pablo Cantero @pablocantero Thank you