File Systems and Schedulers: WAFL and Waffinity and the History of Scaling

Brennan Freeze
NetApp

Stevenson 1300
12:00 PM

Operating Systems such as Linux, Windows, and macOS are general purpose OS that work very well for the average user. However, they are unable to handle enterprise scale throughput of reads and writes at low latency. There are operating systems that specialize in just that, one of them being called ONTAP, a data management operating system that relies on WAFL as its backbone. ONTAP has been used across several sectors as the core way many enterprise level organizations store and manage their data. This talk is to go over what the data storage industry is like and what the filesystem WAFL is. This would entail details of how it works to handle snapshots, consistency points, and protecting data when system failures happen. Reads and writes would not be possible without a file system scheduler. WAFL relies on Waffinity, a file system scheduler that relies on the current structuring of WAFL (Aggregates, Virtual volumes, Private and public file space, etc.) to execute safe reads and writes on the system. Waffinity has gone from first running only on a single core CPU, to now being able to execute on dozens of cores in parallel, maximizing the systems hardware to the best of its ability by organizing itself as a hierarchical file system scheduler.