This would require cwhisper to allow users to send in multiple data points using the same timestamp, and similar to A whisper of many aggregation policies (with percentiles), we might also need a new query notation in carbonapi & go-carbon.
However, a new way of block size estimation is also needed because now we need to take into account of overlapped data points. Some hints might be needed in retention schema to teach cwhisper preallocate a good file size to avoid auto-extension.
And we could also support "overlapped timestamps" for percentiles aggregation, which is almost going to be done in a separate PR.
This would require cwhisper to allow users to send in multiple data points using the same timestamp, and similar to A whisper of many aggregation policies (with percentiles), we might also need a new query notation in carbonapi & go-carbon.
However, a new way of block size estimation is also needed because now we need to take into account of overlapped data points. Some hints might be needed in retention schema to teach cwhisper preallocate a good file size to avoid auto-extension.
And we could also support "overlapped timestamps" for percentiles aggregation, which is almost going to be done in a separate PR.
https://github.com/go-graphite/go-whisper/pull/5 https://github.com/go-graphite/go-whisper/issues/12