CASE STUDY: Oracle to Amazon Aurora Migration Delivers $1M in Savings, Enhanced Reliability

Lee Hylton Amazon Web Services, Case Studies, Cloud Technology, DevOps, Migration

The COVID-19 pandemic caused many businesses to re-evaluate the means by which they backup and protect their digital platforms and mission-critical data. Blue Sentry worked with one such customer – a company that supports a unique platform to provide critical event management, alerting and notification services in a real-time fashion.

Keeping Up with Demand

As the COVID 19 crisis intensified, this company proactively evaluated all of its critical systems for business continuity in order to ensure reliability and performance for all of its customers as demand grew rapidly. They quickly realized that one key application, which relied on Oracle databases in a colocated data center, had a less than desirable business continuity capability. Due to cost-prohibitive Oracle licensing, these databases had no true failover capability, instead relying upon on a backup/restore methodology in the event of catastrophic failure.

While this was a workable solution, it was much less than desirable. In the event of a failure, the RPO was 24 hours to get the application stack back up and running. The company wanted to do better, but was limited both by the capability of its co-located infrastructure and the cost-prohibitive nature of establishing true failover capability using Oracle. Oracle licensing fees would have exceeded $1 million annually to establish a single redundant failover capability. In addition, the company faced a looming multi-million-dollar upgrade requirement on the primary production databases as their Oracle licensing agreement was reaching expiration.

A Cost-Efficient Solution

Blue Sentry worked quickly to analyze the Oracle databases in order to propose an appropriate solution. Upon review, we recommended a full migration to Amazon Aurora, a cloud-native, enterprise-class database solution from AWS. Aurora would cost a fraction of Oracle license renewal for just the current capability, and Aurora offers near-zero RPO and RTO with multiple replicas across multiple availability zones. Aurora also offers multi-region capability, which was attractive to our client. They would benefit from a dramatically enhanced capability for a fraction of the current cost, while avoiding all of the looming expense of an upgrade.

Using the AWS Database Migration Service (DMS), Blue Sentry was able to migrate the large datasets from the on-premise Oracle platform to Aurora quickly and efficiently with a minimum of manual intervention. To keep costs down, the Blue Sentry team wrote a module in Terraform to provision the AWS DMS solution in both testing and staging environments. The migration of the datasets —comprised of millions of rows of data to Aurora —was accomplished in a matter of days. And, because of the infrastructure-as-code approach that Blue Sentry took to the migration plan, these testing and staging environments could be destroyed when testing was completed, saving additional cost. In addition, the infrastructure code included multi-region capability should the company wish to implement that for these databases in the future.

The Result

Our customer dramatically increased the reliability of its data plane, while lowering existing costs at the same time. Also, they gained a more scalable cloud-native solution for hosting the data plane while also avoiding the big upgrade cost on Oracle. Using AWS Aurora for the data plane with automatic failover means they can measure recovery time in seconds instead of days. The migration to Aurora saved over $1 million a year in licensing costs at the current capability level; the company saved an additional amount in excess of $1 million when compared to what it would have cost to achieve the additional capability on collocated Oracle.

Ignoring the additional human costs and related collocation costs, our client conservatively saved over $2 million per year. They also have Terraform Code to make future testing of database enhancements quick and easily migratable through Dev, Test, and Staging environments; plus they retain the modules that Blue Sentry created so that they may deploy into a new environment or implement multi-region capability whenever they wish.