Along the way, we have not only halved the cloud price
but also optimized the infrastructure
and automated deployment processes
. Cost Reduction
The expense reduction process took us several steps to execute a fourfold price cut:
Routine Processes Automation
- Migration of all big servers from Amazon to Azure. This allowed a considerable cost reduction. Later, the client considered the need for additional cost-cutting.
- Server optimization. Our team conducted an in-depth analysis of the existing servers on Azure, which allowed us to optimize the processes. It helped cut the expenses by 40%.
- Migration of the servers from Azure to Oracle. Content Analytics got a special offer from Oracle Corporation for servers hosting with the initial configuration. Since the migration to Oracle, the price for cloud services has dropped fourfold.
To test a new feature CA team had to clone new production servers which became extremely time-consuming. Before migration to Kubernetes
, it took CA about 20 minutes to create a new environment for a client.
OpsWorks Co. helped CA automate copying, creating, removing, and migration of servers. Now, creating a new production environment for a client requires pressing a single button in their CI/CD tool. Once automation has been executed, it takes 20 seconds maximum.
Content Analytics was already using Jenkins
as a Continuous Integration tool for its infrastructure. Cloud-based Ecosystem Construction
There are hundreds of servers to host Content Analytics infrastructure. Previously, the CA team used OpsWorks
for the servers management and configuration. As our primary goal was to reduce costs, we decided to migrate the CA infrastructure to Kubernetes, which consequently simplified the management of the servers.Deployment Optimization
As it was mentioned before, Content Analytics owns hundreds of servers. When a feature was tested, engineers deployed code on tens of servers simultaneously. If one of them failed, the deployment stopped, and they had to launch the process manually.
We used Jenkins to optimize and advance the infrastructure. With this technology, test environments are created and killed due to emerging needs.Monitoring Flexibility
Initially, Content Analytics was using Nagios
as a monitoring system for its infrastructure. Nagios approach had several disadvantages. It involved manual work for severs creation and removing. The monitoring solution developed by our engineers was based on Prometheus
integration. We use Prometheus for clusters monitoring and Zabbix as an auxiliary tool. Also, it involved ELK Stack implementation for logging.
Zabbix analyzes the system work and predicts possible issues. Also, any new server that has been created automatically appears in the monitoring system. Therefore, Prometheus and Zabbix setup gave us an opportunity to react quickly and solve issues more effectively.Project Results
1. Deployment time reduction from 30 to 5 minutes
by virtue of optimizing and improving deployment scripts and processes
2. Fourfold costs reduction
due to databases migration from RDS to Azure
3. No routine job
owing to automated on the production and testing stages
4. Shorter time-to-market and improved customer satisfaction
due to automating and optimizing CI/CD Pipeline.