EZOPS Reduces Delivery Time for POC Environments to 1 Day

Case Study

EZOPS Reduces Delivery Time for POC Environments to 1 Day

Executive Summary

Risk Focus, a Ness Company, conducted a Well-Architected Review for EZOPS and determined its could reduce its service-level agreement (SLA) to build proof-of-concept (POC) environments by adopting best practices to create Infrastructure as Code. The end result was to reduce the SLA from the DevOps team to deliver POC environments from about 4 business days to 1 and to fully automate the process, thereby ensuring no manual errors.

About EZOPS

EZOPS Inc. is a U.S.-based FinTech firm providing full front-to-back data control software that drives efficiency and dramatically reduces operational costs and resource requirements for financial services institutions.

With the power of its novel machine learning tools capable of automating virtually any manual process, EZOPS has harnessed artificial intelligence to vastly enhance data flow and reduce operational bottlenecks, enabling clients to enjoy major cost and time savings while achieving straight through processing (STP) automation goals.

The Challenge

A fast growing AI startup, EZOPS found that the time to build and verify secure “sandbox” environments to work with clients was taking too long to create and validate. The company sought options to improve the turnaround time and to keep up with demand.

 

Why AWS

EZOPS chose AWS for the project so it could leverage the elastic nature of the cloud by spinning up sandboxes on demand. They liked AWS because it provided the ability to create secure, encrypted environments for each client that could be completely destroyed at the end of the POC, ensuring the full protection of all client data. From a cost perspective, EZOPS saved considerably over the previous “hosted” infrastructure it had used, because the company could limit running it to when it needed the environments and then destroy it at the end of the POC with no ongoing costs.

Why EZOPS Chose Risk Focus

EZOPS chose Risk Focus to do a Well-Architected Review based on recommendations from mutual clients and because Risk Focus is an AWS Advanced Partner with AWS DevOps and Financial Services Competencies.

The Solution

To address the recommendations of the Well-Architected Review while keeping costs to a minimum, EZOPS asked Risk Focus to provide a solution that could be delivered within a budget of 10 days’ effort. The solution focused on the Amazon Machine Image (AMI) construction and core AWS build, since these are the most fragile and time-consuming part of any deployment. EZOPS elected to defer automation of the account setup and VPN setup in order to focus on the most time-consuming and complex steps.

Risk Focus migrated the manual process of building an environment to Infrastructure as Code, automating 72 manual steps by creating a 2-step process to install a AWS CodeBuild job into the AWS account and to then run CodeBuild to build the infrastructure with HashiCorp Terraform. The solution had 2 parts:

Part 1 – Automation of the AMI build and database setup. This was solved by introducing the open-source software HashiCorp Packer to build the AMI and leverage Terraform to generate database passwords dynamically when CodeBuild runs Terraform. The database setup was handled by leveraging the user data script in EC2 to initialize the database after the EC2 is created.

Part 2 – Convert the manual process to create 27 AWS entities (VPC, EC2, RDS, etc.) to Infrastructure as Code. This was delivered by creating a CodeBuild project installed using the Terraform command line. The CodeBuild project is then run with configurable parameters to complete the infrastructure build with the push of a button.

Results and Benefits

Risk Focus delivered the project in less than 2 weeks. EZOPS can now create an environment in less than 1 hour that used to take about 4 days, and it can now scale the reproducible process to staff in multiple locations. In addition, every deployment is fully auditable, because all provisioning is stored as code.