In February of 1888, John Boyd Dunlop’s son had a heavy cold. His doctor prescribed a very unusual cure: cycling. Dunlop bought the child a tricycle, but there was a problem. These were the time of cobbled streets in Scotland. His son complained of having headaches while riding the cycle because of the uneven terrain and hard tires. This gave Dunlop the idea of using an inflatable rubber tube to cushion the ride. While hardly on the same scale, Delta Bravo was created to solve a business problem that’s given IT headaches for years.
The Gift, The Curse, The Pivot
Delta Bravo actually started out as a service company, focused on building artificial intelligence (AI) and machine learning (ML) applications for others. After completing a large scale AI platform for a global client, we were asked to assist in ongoing management and development. We jumped on the opportunity but drastically underestimated what was involved. Our client wanted us to:
- Manage performance for thousands of SQL Server instances with versions ranging from on premise 2008 R2 up to Azure database
- Secure databases up to Department of Defense STIG standards
- Provide estimates and predictions for how data growth would impact their monthly cloud costs
- Identify target workloads to move to the cloud
- Engage in CI/CD process for application development and release
The Delta Bravo “A-ha!” Moment
The tools and techniques for relational database management haven’t changed much in the last 15 years. Our small team struggled with our client’s demands at first, vacillating between several different tools for monitoring, performance improvement and capacity planning- not to mention the DevOps aspect of rolling out capabilities in a fluid manner. We realized that database security was still a largely manual process frought with risk. We were afraid the everyday “care and feeding” of the environment would overwhelm us; there was so much to do, many days we didn’t know what to do first.
We decided to tackle this challenge by doing what we did best: build a world-class Artificial Intelligence platform to solve the problem- and Delta Bravo AI was born.
Using AI to handle Performance, App Dev, Support and Security
We didn’t set out to solve our problem by building a me-too version of an existing product or a “mash up” of all of the tooling we thought we needed. We forced ourselves to answer whether or not data science, machine learning and artificial intelligence could:
- Predict performance and security issues 30 days in advance, quantify their impact and offer the best possible solution for remediation
- Prioritize optimizations based on impact
- Find anomalies in performance and security, tie them to specific events
- Apply database performance and security solutions against multiple models to determine the best solution on a database-by-database basis
- Quantify and predict impact of code changes prior to deployment in production
- Analyze database performance and security at scale, handling thousands of inputs without impacting performance
We didn’t approach this problem like DBAs, we approached it as Stewards of the Data- the concept being that we needed to create value for each group impacted by the data, from IT Ops to Developers to Security Teams to CFOs and CIOs. Our vision was far broader than keeping the database “up and running.”
Mistakes are the Stepping Stones to Wisdom
Building Delta Bravo to solve the problems we had was an ambitious undertaking. We evaluated several development languages, eventually settling on Go. There was a lot of testing and thought put into agent vs. agentless data collection. We are the first platform on the market to use predictive analytics on database performance and growth, which is very difficult to build relevant, repeatable models for. There was a LOT of trial and error and many lessons learned, impacting everything from our feature set to performance to balancing the cost of doing all of this in the cloud.
Today, we speak at public events about these lessons learned and work with with several larger companies who glean our advice on building production-scale AI platforms.
AI empowers a whole new perspective
Whether you’re already using AI or considering it, we hope our story helps you continue in that direction. Database monitoring and management tools have been around for 15 years. With AI, it took us two years to leapfrog all of them with a platform that solves a business case, not just an immediate technical need.