SurveyMonkey
Staff Database Engineer
SurveyMonkey is a global leader in online surveys and forms that empowers people with the insights they need to make decisions with speed and confidence. Our fast, intuitive feedback management platform connects millions of users worldwide with real-time AI-powered insights that drive meaningful decisions. We provide answers to more than 20 million questions every day so that people and organizations can attract new audiences, delight customers, create advocates, and extend their competitive advantage in the marketplace. Our vision is to raise the bar for human experiences by amplifying individual voices. Learn more at surveymonkey.com. Learn more at surveymonkey.com.
- In this role, I am part of a team with 3 other Senior Database Engineers that maintain multiple multi-Terabyte databases that exist within Aurora MySQL and SQL Server. They are setup respectively as a cluster and within availability groups. We maintain our database systems with zdt and ensure the scalability of our databases. We do all our maintenance online, only allowing only minutes of downtime once per quarter. We utilize Terraform, Powershell, Python, and Lambda functions to automate as much as possible. We also use SQL Sentry and Cloudwatch for monitoring and alerting to keep us in touch with the health of our databases.
- I set the direction for best practices and have led multiple cross-functional projects including our migration from PYMSSQL to PYODBC. I am also responsible for capacity planning and performance management. Since I have been here, we increased the durability of our data files from 96.6% durability to 99% durability at minimal cost as well as increased our durability of our T-Log files to 99.999% durability, resulting a much-needed stability of our database systems.
- I have done significant analysis of our systems to "Rightsize" EC2 and EBS volumes. We were able to get significant savings by making adjustments. We also completed a plan in 2023 that resulted in halving the number of vCPUs and at the same time doubling the amount RAM at no added cost. This also resulted in significant license savings.
- I have also tuned dozens of high resource consumption queries. The most prominent was called over 200 times a minute and was able to decrease the CPU Time on that sproc by 99%, the average reads by 99% and the average runtime by 40%. The next was a sproc that is called over 1200 times a minute. It’s runtime was reduced by over 50% and we saw a 60% reduction in CPU usage per execution, which equated to a 2% drop of the total server's CPU usage.
- We have also migrated multiple MySQL Clusters into Aurora and have a few more to go.