Blog: why histograms? #2658
Blog: why histograms? #2658
Conversation
|
FYI - formatting check failed, but since we're probably going to do copy-edits, let's minimize your pain and only have you reformat stuff until after we're satisfied with the content 🙂 |
It's just as easy to do it now. Reviews may be easier if things are in the expected format. |
There was a problem hiding this comment.
Looks great overall, thank you! Mostly some small copy-edits, but I modified the last paragraph a a bit to caveat about backends and sampling strategies. Since accurate calculations of SLOs is a super deep topic we probably shouldn't elaborate too much, but I do think we should call out that the problems you can run into won't be the same from backend to backend.
|
updated the date to monday may 8th to reflect when this is likely to be merged |
|
If a maintainer merges this before that obviously they can change the date. I have enabled "allow edits from maintainers" on the PR |
|
@cartermp @chalin I rewrote the section "Other metric types" because on re-reading I noticed that the wording was technically incorrect. The last paragraph previously stated that collecting each percentile as a gauge is an ad-hoc implementation of histograms, when this is actually much closer to a summary metric type. I also added a caveat that summaries cannot be aggregated across hosts. PTAL |
Co-authored-by: Phillip Carter <pcarter@fastmail.com>
Co-authored-by: Phillip Carter <pcarter@fastmail.com>
|
🎉 |
Guest post from https://dyladan.me/histograms/2023/05/02/why-histograms/
This is the first in a series of 3 posts. The second explains the differences and tradeoffs between histograms and summaries (https://dyladan.me/histograms/2023/05/03/histograms-vs-summaries/). The third explains exponential histograms in detail (https://dyladan.me/histograms/2023/05/04/exponential-histograms/).
Please feel free to share any editing or copy suggestions.
Preview: https://deploy-preview-2658--opentelemetry.netlify.app/blog/2023/why-histograms/