Articles by Will Larson
An Elegant Puzzle: Systems of Engineering ManagementWill shares his view of engineering management as a series of elegant, rewarding, and important puzzles, and advice for how to solve them.
- Notes on data modeling from Handbook of Relational Database Design
Will shares some notes from reading the first half of Handbook of Relational Database Design. He reviews the three-schema approach, properties of an effective data model, taking a data-driven approach towards modeling data, and essential vocabulary.
- Metrics for the unmeasurable.
Will discusses the importance of metrics in aligning complex organizations, and how to think about defining metrics in security, productivity, and other areas that defy measurement.
- What do Staff engineers actually do?
Will's guide to being a staff engineer covers: • Setting technical direction • Mentorship and sponsorship • Providing engineering perspective • Exploration • Being Glue
- Cold sourcing: hire someone you don't know.
Will covers: • Why to move beyond your personal network • Your first cold sourcing recipe • Is this high-leverage work?
How to present to executivesWill introduces tips on communicating effectively and mistakes to avoid when presenting to executives.
- Build versus buy.
When deciding whether to build a solution internally or buy from a vendor, companies should consider risk, value, and cost. Risks include the vendor going out of business or changing their offering. Value comes from how the vendor's solution compares to what you can build, both now and over time. Costs include integration, financial, operating, and evolution costs over the lifetime of using the vendor. Ultimately, companies should determine if any risks are unacceptable, then perform a simple value versus cost calculation to make the decision. However, companies often neglect vendor management, which is key to extracting full value from vendors.
- Introduction to architecting systems for scale
Will documents some of the scalability architecture lessons he's learned while working on systems at Yahoo! and Digg, around load balancing, caching, off-line processing, and a platform layer.
- Your first 90 days as CTO or VP Engineering.
Will covers: • VPE vs CTO • Priorities and goals • Making the right system changes • Tasks for your first 90 days across learning and building trust, creating a support system, organizational health and process, hiring, execution, and technology
- Don't follow the sun
Will argues that on-call should not split into multiple shifts within a 24-hour window, with reasons that include: • Shifts start to turn into service windows, as opposed to exception management • Shift transitions magnify error rate • Humans are too slow, anyway; the focus should be on automation that prevents exceptions