fractional_sample_rate: Vary sample rate according to traffic estimation #4839
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…past 30 days Our old sampling mechanism used SAMPLE 20000000 syntax. This was wonderful since it allowed essentially dynamic sampling based on the data being queried. However this ran into many issues relating to JOINs and sample rate being different for different tables. Instead, we now start to dynamically vary sample rates fractionally. At query time we check the time window being queried and estimate how many rows this query might reach. For large queries, we then dynamically decide the sample rate. For getting the traffic estimate for a site, we have a new SampingCache class which queries `ingest_counters`. The query being cached is slightly expensive and can be sped up with a ClickHouse projection.
macobo
force-pushed
the
sampling-rate-changes
branch
from
7b7982b to
7ee1254
Compare
macobo
force-pushed
the
sampling-rate-changes
branch
from
3694063 to
f7e8e3a
Compare
macobo
added a commit
that referenced
this pull request
Nov 20, 2024
Companion to #4839 A projection speeds up ClickHouse queries by storing data in a different order in each part. In this case, it pre-aggregates results by site, speeding up the query and requiring less resources at query-time. Query stats before: ``` -- 10 rows in set. Elapsed: 1.134 sec. Processed 1.45 billion rows, 21.65 GB (1.28 billion rows/s., 19.09 GB/s.) -- Peak memory usage: 53.41 MiB. ``` After: ``` -- 10 rows in set. Elapsed: 0.321 sec. Processed 39.70 million rows, 1.07 GB (123.51 million rows/s., 3.33 GB/s.) -- Peak memory usage: 97.43 MiB. ```
aerosol
approved these changes
Nov 20, 2024
Co-authored-by: hq1 <[email protected]>
aerosol
reviewed
Nov 20, 2024
macobo
force-pushed
the
sampling-rate-changes
branch
from
f68928d to
db34e53
Compare
github-merge-queue bot
pushed a commit
that referenced
this pull request
Jan 7, 2025
* Speed up counting past 30 day traffic via ingest_counters Companion to #4839 A projection speeds up ClickHouse queries by storing data in a different order in each part. In this case, it pre-aggregates results by site, speeding up the query and requiring less resources at query-time. Query stats before: ``` -- 10 rows in set. Elapsed: 1.134 sec. Processed 1.45 billion rows, 21.65 GB (1.28 billion rows/s., 19.09 GB/s.) -- Peak memory usage: 53.41 MiB. ``` After: ``` -- 10 rows in set. Elapsed: 0.321 sec. Processed 39.70 million rows, 1.07 GB (123.51 million rows/s., 3.33 GB/s.) -- Peak memory usage: 97.43 MiB. ``` * deduplicate_merge_projection_mode * materialize projection
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Our old sampling mechanism used SAMPLE 20000000 syntax. This was wonderful since it allowed essentially dynamic sampling based on the data being queried. However this ran into many issues relating to JOINs and sample rate being different for different tables.
Instead, we now start to dynamically vary sample rates fractionally.
At query time we check the time window being queried and estimate how many rows this query might reach. For large queries, we then dynamically decide the sample rate.
For getting the traffic estimate for a site, we have a new SampingCache class which queries
ingest_counters.Future work:
query.filtersis set.