A Journey to AWS Cloud Continuous Cost Optimization (CCO)

Read Time 4 mins | Written by: ASCENDING Admin

Background:

Tool

From 2014 FrugalOps team dedicated to Amazon Web Service (AWS) Cloud Continuous Cost Optimization (CCO), we have built our own cost monitoring and optimization tool FrugalBot for AWS Cloud and distributed on AWS Marketplace. However, we felt something is missing other than a tool in the journey of staying cost optimum on AWS since new services and concepts are coming out everything literally.

Many enterprise were conscious about cost optimization in the second year in the Cloud. By the time you “Lift and Shift” your architecture onto AWS, it may be already late. We constantly think the best way for folks like you to cut AWS cost in any stage.

Method

When the team came back from 2016 Re:Invent, we heard lots of buzz word like spot instances, autoscaling, stateless, serverless, caching etc in terms of cost optimization.  We realized there were so much existing tricks and best practice we can share to folks. Perhaps it will be useful for you if we pick some critical pieces and blend in practical example around those core AWS cost saving techniques.

We wish everybody can be beneficial from the combination of the tool and method.

The Term – Continuous Cost Optimization:

“Pay as you go”

It would be good to have a tool in place or even Amazon built in billing system to help your cost allocation and reserve instances(RI) planning etc, but that’s not end of it. The “pay as you go” cloud is a double-edged sword. It can kill the devil , also it can summon the devil. We have seen company like Mapbox running 90% of production workload in spot instances. We’ve also seen a client used to go through painful RI purchase process each month, ran all production loads in a set of static servers. Just for the apple to apple comparison, Mapbox run 100 servers vs folks run 100 servers in RI or on-demand per day, they only pay  20% cost of later one.

We only build “Skynet”

So even you have the best tool in this field, it’s still hard to manage cloud cost if you didn’t build cost awareness into team’s culture. AWS call it “Lift and Shift”. AWS offer so many components and services to architecture modern stack in the cloud effortlessly , it’s a big question what to use other than EC2 instances. In our AWSMeter team, we consider building a new system on cloud is like building a Skynet who can endlessly reproduce and scale up/down anytime, anywhere. Don’t worry unlike movie we are not “Cyberdyne”, I am not Miles Dyson. lol . However, we think everybody can build Skynet grade app on cloud, thanks to Amazon for pushing us to the era.

This is why we are so obsessed with cost optimization architecture on cloud, not only it can bring you financial benefit but also make your application dynamic and false tolerance.

Continuous Cost Optimization

Continuously evolving is the key on AWS. AWS has and will rapidly release new components and services, you can checkout 2016 re:invent keynote from Andy Jassy. Looking at the number of features they have released in 2016 alone, we felt so excited and challenging. Your architecture has to adopt those changes quickly to be more powerful and cost efficient. Over the last few years, we saw all of our clients increasing their computing loads day by day. Continuous Cost Optimization (CCO) and architecture evolving will be the key to your successful business expansion. It’s also our team’s day to day mission to help you get on the trend and bring more benefit to your organization!

Tackle:

The best way to share those tricks on cost optimization with real life code example and tools to help you achieve CCO in your organization. We decided to tackle some of pieces in next few months.

  1. Spot instance and SpotFleet – We have converted some real life app to run on instances in spot market cutting about 80% total cost.
  2. AWS components and services – By using fully utilizing AWS components and service, you can achieve some quick saving.
  3. Caching – Big company are caching everywhere and anywhere they could. It can tremendously cut down throughput and computing power.
  4. Database – No fighting on relational or noSql database! The answer should be both in AWS.
  5. Tools and measurement to CCO – Building a habit of measuring cost and optimize it.

Commitment!

It was about 5 years ago that we started pet project on AWS. We have seen rapid improvement and tremendous opportunity in it. We tried many tools and approaches to manage cost for our clients but we still saw lots of folks are struggling for the cost in the Cloud.  Hopefully, we can spread the concept of “Continuous Cost Optimization” into community through those examples. Our entire team are so excited for next 5 years and looking forward to work with you in the ecosystem. Please feel free to leave your comments on any of our post. We are pleased to talk to you if you need any help.

 

Best,
FrugalOps Team
Email: blog@frugalops.com
Twitter: @frugalops

 

 

ASCENDING Admin