arrow_back How resilience testing of infrastructure is the path to achieve stable,fault-tolerant and self healing infrastructure
How to Create a Low Cost, Self Healing & Immutable Infrastructure using AWS EC2 Spot Instances
Submitted by Aditya Upadhyay (@aditya-u) on Friday, 10 April 2015
Reduce AWS bills significantly, automate maintenance using Auto Scaling & understanding spot instance failover
AWS is a leading cloud services provider and it has millions of users. The major component of
AWS usage & subsequently billing are EC2 Instances. Although the cost of EC2 instances goes
down if one purchases reserved instances, there is another feature for EC2 instances which is
rarely used by customers. This session will guide you on how to use it.
AWS provides spot instances which are the unused compute resource of AWS infrastructure &
can available to customers at a very low price. Prices are governed by market dynamics of
demand & supply with users putting up bids to request these instances. Average market prices
can be 1/8th to 1/10th of normal instances, and when users’ bid prices exceed market prices
these requests are fulfilled.
Only catch is, AWS can terminate allotted instances with little to no notice if A) market price
exceeds bid price or if B) the spot pool of your instance type is reduced due to increase of
demand for normal instances.
These are the risks why many users stay away from Spot instances. This session will provide a
solution to address these risks which actually reduces costs without compromising performance.
The solution includes deploying AutoScaling Groups setting attributes like Launch
Configuration, VPC & Subnet, Number & Composition of instances, Notification settings and
This session will have a demonstration of deployment of AutoScaling groups & attributes,
Launch of instances and simulated termination to display its Self Healing capability. I’ll also discuss two
real situations in detail.
Having managed DevOps for various clients, deploying this solution has produced positive
results of reduced AWS bills & infrastructures capable of healing themselves.
I am a Senior Consultant at CodeIgnition and manage DevOps, Cloud & Continuous Delivery projects. With 3 years of hands on experience & many more as a code connoisseur, I am passionate about Continuous delivery, Cloud, Infrastructure & Automation. I gave a 15 minutes talk at Rubyconfindia, about a very similar topic. This talk will be a detailed version of that talk. I am using all the concepts, I wanna talk about, in the architecture of a continuous delivery platform I am developing, greenci.io.