944115 - Estimate size of Telemetry payloads with UI Telemetry

Status: RESOLVED WORKSFORME

(Firefox :: General, defect)

Description

[Vladan Djeric (:vladan)]

This isn't really a bug, it's more of a glorified "needinfo".

Can I get an estimate of the amount of data collected by UI Telemetry for a long-running session?

If we discover that UI Telemetry will be doubling or tripling the size of Telemetry payloads, we should tweak the UI event notation to be more compact and less repetitive.

Comment 1

[Mike Conley (:mconley) (:⚙️)]

(In reply to Vladan Djeric (:vladan) from comment #0)
> we should tweak the UI event notation to be more compact
> and less repetitive.

One way to do that is to maintain our own internal count of how many times an event occurs as opposed to logging each event. This would mean not using UITelemetry's built-in event logging mechanism. That'd also sidestep the potential privacy issue brought up by bug 944116, since we wouldn't record timestamps at all.

Comment 2

[Mike Conley (:mconley) (:⚙️)]

(In reply to Vladan Djeric (:vladan) from comment #0)
> This isn't really a bug, it's more of a glorified "needinfo".
> 
> Can I get an estimate of the amount of data collected by UI Telemetry for a
> long-running session?
> 

Let's suppose that a long running session is 30 days.
Let's suppose that each day, a user triggers 500 events that are logged.
Let's suppose that each log is approximately 100 bytes long.

This equals an upper bound of 1.5MB.

That is quite high. :/

Comment 3

[Mike Conley (:mconley) (:⚙️)]

(In reply to Mike Conley (:mconley) from comment #2)
> Let's suppose that each log is approximately 100 bytes long.

I should have said, "Let's suppose that each event log is 100 bytes long"

Comment 4

[Mike Conley (:mconley) (:⚙️)]

With bug 945499, we have a much nicer upper-bound on packet size. I don't think this bug is a concern anymore.