Optimized date parsing for numbers #132462
Conversation
felixbarny
added
>non-issue
:Search Foundations/Mapping
elasticsearchmachine
added
external-contributor
|
Pinging @elastic/es-storage-engine (Team:StorageEngine) |
|
Pinging @elastic/es-search-foundations (Team:Search Foundations) |
… as a number This improves the ingestion performance in Elasticsearch as it can leverage an optimized code path for date parsing. See elastic/elasticsearch#132462 This is not a breaking change because we are using the `date` field type for metrics. Supporting a more granular date format, such as `date_nanos` doesn't seem sensible for metrics as it would cause a big storage overhead. Most other metrics data stores also don't support more granularity than milliseconds.
|
Well, today, the Elasticsearch exporter in the OTel collector sets an epoch millis timestamp with a fractional for the But for metrics in particular, I don't think this level of granularity is useful and we're using the |
… as a number (#41811) This improves the ingestion performance in Elasticsearch as it can leverage an optimized code path for date parsing. See elastic/elasticsearch#132462 This is not a breaking change because we are using the `date` field type for metrics. Supporting a more granular date format, such as `date_nanos` doesn't seem sensible for metrics as it would cause a big storage overhead. Most other metrics data stores also don't support more granularity than milliseconds. --------- Co-authored-by: Carson Ip <[email protected]>
3 participants
Optimized parsing for
dateanddate_nanosif the data is provided as epoch millis.