Surfline started using MongoDB back when they were running in a physical data center. A few years ago, they moved to Amazon Web Services, but continued self-managing their MongoDB deployment. Matt Walker, Senior Engineering Manager at Surfline, recalls how they maintained their servers. “When running MongoDB ourselves, we built a lot of tooling to support backups, to manage users, and to monitor MongoDB clusters.”
For examples, backups were taken with mongodump, managed by a series of Python scripts, and pushed to Amazon S3. Not only was this process error-prone, but it also impacted the performance of their production database.
As the company grew, the engineering team felt the burden and inefficiency of managing their own tooling and Gavin Cooper, Senior Director of Product and Engineering, knew it was time to make a change. “Over time, we also realized that what we had originally provisioned was may more than what we needed. And at that point, we could have either reprovisioned a cluster to again manage ourselves or just migrate our data to MongoDB Atlas and never have to worry about this problem ever again.” After the move, a weight was lifted off the team.