Combined Dynamic Sampling #111
Conversation
|
it looks good, but what i miss from RFCs often is some kind of diagram to help visualize what we are talking about. then something more like a use case that is a concrete user example: sub trace for example I don't really understand what this means. Sampling on the individual events? There is no trace? |
|
@JoshFerge This RFC is not yet ready for review, it's still in draft state. |
| the sample rates permit, and dynamic sampling will prefer to store traces with | ||
| all data available. | ||
|
|
||
| ## Sub-Trace Sampling |
There was a problem hiding this comment.
This section will be revisited. Following a conversation with @mydea, we will keep the replay sample rate completely independent of trace sampling (however subject to RFC 109). I'll write up a revision of this section that takes necessary changes into account.
got it. sorry about that. it's an exciting one :D |
Reduce the number of profiles and replays dropped through dynamic sampling by
changing how trace sampling decisions are computed.
Rendered RFC