Provide multiple default analysis layouts (for different problem scenarios)

[ebugden]

Providing several good default layouts for different common trace analysis problems (ex. debugging a latency) or contexts (ex. network tracing) helps new users adopt good analysis methodology and solve problems faster.

For example, when a user opens a trace they could specify the context or goal of the tracing session and the default views would be adjusted based on this.

---

Good default layouts are:

• Based on observation of analysis experts: Do different analysis experts use similar approaches for similar tracing problems? Which analysis steps are common to all problem types? Implementing #267 could help answer these questions.
  • Common step example: Verifying that you opened the correct trace (Does the metadata match the expected system, time, etc.? Is the problem to be analyzed present in the trace?)
• Start with high-level overviews: #393 Starting the user in a place that feels familiar (ex. system call level) orients them before they dive deeper
• Conservative (rather than complete): Problems that require tracing to solve are specialized so it's impossible to include every analysis that could be useful. By keeping default layouts minimal (only including views that are useful to everyone) we avoid annoying the experts with clutter and we encourage new users to make choices and actively engage in the analysis process.

---

CC: @ssmagula

[ebugden]

Maybe Trace Compass Perspectives (View layouts) could inform these default views?

[ebugden]

Some high-level overview view ideas:

• CPU top
• Strace
• Application level
• Active thread
• Histogram that distinguishes between event types