You are here
Everything an Administrator Wanted to Know about Operating System Internals but Was Afraid to Ask
Thurgood Marshall South
The Morning Session (Part 1) of this tutorial is FULL.
Insufficient knowledge of operating system internals is my most common reason for passing on an interview candidate. Anyone can learn that you run tool X to fix problem Y. But what happens when there is no tool X, or when you can't even accurately pinpoint the root cause of why "It's sometimes slow."
This will be a no-holds-barred, fury-road-paced review of all major parts of modern operating systems with specific emphasis on what's important for system administrators. It will provide just enough of an academic focus to bridge the "whys" so you can make better use of fiddling with the "whats" on a day-to-day basis. As an added bonus, it will prime you for the following day's "Linux Performance Tuning" tutorial with Theodore Ts'o.
You will learn about process management, scheduling, file system architecture and internals, interrupt management, the mysteries of the MMU and TLB, belady's anomaly, page replacement algorithms and hopefully a bit of networking. In a nutshell, we'll cover 16 weeks of college-level material in a few hours.
- All admins who did not take the Comp-Sci academic route and never had a course in OS internals
- Inexperienced admins whose coursework or training didn't include the kind of OS internals that they should (modern OS courses have become a shadow of their former selves and commonly require writing no OS code)
- More experienced admins who haven't really had to address these sorts of issues on a regular basis who probably know a lot about some individual aspects but could benefit from everything being put into a broader context
Attendees will gain a deeper understanding of what goes on inside the kernel and the areas where things can go wrong. We'll explore how little the concept of "system load" captures about the true system state, and attendees will be prepared to improve both their operational response methodologies as well as their monitoring goals.
- Scheduling and Process Management
- Memory Management and the MMU
- Virtualization and its impact on these
- File System Architecture (for sysadmins covering ext (2, 3, 4), NTFS, and ZFS)
- Storage layer performance, disks, RAID, and SANs
- The impact of virtualization on these
All attendees should have either a laptop running Linux, or a VM with a recent copy of Linux installed to assist in the hands-on exploration activities.