This past fall, I had the opportunity to intern on the relatively new Bulk I/O team at Cockroach Labs. Our team’s focus is to improve bulk data operations on CockroachDB. My work involved creating a better experience when importing or backing up data without using an external storage service like Amazon’s S3 or Azure’s Blob Service. CockroachDB supports interacting with files in a node’s local storage system, through a feature called nodelocal. However, this has caused confusion and misuse due to its implementation, and the project I took on involved refining the user experience around nodelocal.