If You Can't Draw the Boundary, You Don't Have an Architecture
The boundary question ended my whiteboard round before it started. The interviewer asked where my system ended. I listed components. He said I had just described the whole company. He was right. Watch the 40-second version first: Q2: System Design Interviews Start With the Boundary Your architecture starts at the word no Interviewers ask "where does your system end" because that is where the…
In a system design interview, the first question is often about the boundaries of the system. This is because the interviewer wants to understand where the candidate's thinking begins and ends. A list of components is not a design, but a shopping list. The real architectural decisions lie at the seams - the points where ownership changes.
If a candidate can't draw the boundary, they don't have an architecture. Most candidates can name the technical boundaries like authentication, payments, and the CDN, but they often forget the most important part - who owns the other side of the line. The team boundary is where real projects fail, not the technical boundary. The boundary that kills projects is the team boundary, not the technical one.
If you can't name the owner on the other side of the line, you haven't truly drawn a boundary. Interviewers are looking for two things: whether the candidate knows the difference between what they build and what they depend on, and whether they can state the ownership and contracts at each seam. The 4-seam checklist - authentication, payments, data you don't own, and the shell - is a useful tool to remember.
Once you draw the boundary, everything outside gets a name and an owner, while everything inside gets your design. The most important rule is to get the boundary right before you start designing.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
