Skip to content

ClickHouse usage analytics: events/gauges tables with daily MV#3

Open
lohanidamodar wants to merge 97 commits intomainfrom
claude/rebuild-analytics-clickhouse-OHWGZ
Open

ClickHouse usage analytics: events/gauges tables with daily MV#3
lohanidamodar wants to merge 97 commits intomainfrom
claude/rebuild-analytics-clickhouse-OHWGZ

Conversation

@lohanidamodar
Copy link
Copy Markdown
Contributor

@lohanidamodar lohanidamodar commented Mar 14, 2026

Summary

Complete rewrite of the usage analytics library with a two-table architecture optimized for both real-time analytics and billing.

Architecture

  • Events table (MergeTree) — raw request events with dedicated columns for path, method, status, resource, resourceId, country (LowCardinality), userAgent
  • Gauges table (MergeTree) — simple resource snapshots (metric, value, time, tags)
  • Daily MV (SummingMergeTree) — pre-aggregates events by metric + tenant + day for fast billing
  • No periods — query-time aggregation via toStartOfHour/toStartOfDay instead of write-time fan-out
  • Single write path — collect(metric, value, type, tags) routes to correct table; event columns auto-extracted from tags

Key Changes

  • Two separate tables instead of one — events have 7 extra columns gauges don't need
  • Plain MergeTree for both tables — raw appends, query-time aggregation
  • Daily MV with minimal schema (metric, value, time, tenant)
  • LowCardinality(String) for country column
  • Bloom filter indexes on all filterable columns
  • String tenant — setTenant(?string)
  • Utopia Query for all read operations — parameterized queries, no SQL injection risk

API

Write

  • collect(metric, value, type, tags) — buffer with auto-flush
  • addBatch(metrics, type) — direct batch insert
  • flush() — write buffered metrics

Read

  • find(queries, type) / count(queries, type) / sum(queries, attr, type)
  • getTotal(metric, queries, type) — SUM for events, argMax for gauges
  • getTotalBatch(metrics, queries, type) — batch totals
  • getTimeSeries(metrics, interval, start, end, queries, zeroFill, type)

Billing (Daily MV)

  • findDaily(queries) / sumDaily(queries) / sumDailyBatch(metrics, queries)

Test Plan

  • Unit tests for Metric schema, getters, validation
  • Integration tests for ClickHouse and Database adapters
  • PHPStan level max passing
  • Linter passing
  • Security audit — no SQL injection vulnerabilities
  • CI green (CodeQL, Tests, Linter)

- Database adapter
- ClickHouse adapter
- Removed hardcoded column definitions in Usage class, replacing with dynamic schema derived from SQL adapter.
- Introduced new Query class for building ClickHouse queries with fluent interface.
- Added support for advanced query operations including find and count methods.
- Enhanced error handling and SQL injection prevention mechanisms.
- Created comprehensive usage guide for ClickHouse adapter.
- Added unit tests for Query class to ensure functionality and robustness.
- Maintained backward compatibility with existing methods while improving overall architecture.
Comment thread src/Usage/Adapter/ClickHouse.php Outdated
lohanidamodar and others added 14 commits April 8, 2026 05:26
- Remove $useFinal property and setUseFinal() (MergeTree doesn't support FINAL)
- Remove buildTableReference $useFinal param
- Fix resolveTenantFromMetric mixed type handling
- Remove unreachable branch in Database::getTotal()
- Remove always-true count check in addBatch

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Remove separate billing table and MV (monthly aggregation)
- Daily MV uses same column definitions as source table
- Billing queries use daily table (SUM over daily aggregated rows)
- Only events are pre-aggregated; gauges query raw table

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Split the single MergeTree table into two separate tables:
- Events table with dedicated columns for path, method, status, resource, resourceId
- Gauges table with simple metric/value/time/tags schema

Event-specific columns are automatically extracted from tags during addBatch.
The daily SummingMergeTree MV now aggregates by metric, resource, resourceId.
All read methods accept an optional $type parameter to target specific tables,
with null querying both tables transparently.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Query the pre-aggregated daily SummingMergeTree table for fast
billing/analytics instead of scanning raw events.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- country: LowCardinality(Nullable(String)) for efficient low-cardinality storage
- userAgent: Nullable(String) with bloom filter index
- Both extracted from tags into dedicated columns like other event fields
- Added getCountry() and getUserAgent() getters on Metric

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Daily table only has metric, value, time, resource, resourceId, tenant.
No path/status/userAgent/country/tags — those don't aggregate meaningfully.
MV groups by metric, resource, resourceId, tenant, day.
ORDER BY includes resource and resourceId for efficient billing queries.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Use Metric::EVENT_COLUMNS to extract all event columns from tags
instead of hardcoding the list. Now country and userAgent are properly
stored in dedicated columns instead of being left in tags JSON.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Daily table now only has metric, value, time, tenant.
One row per metric per project per day — optimal for billing.
Resource-level breakdown queries the raw events table directly.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Single query with GROUP BY metric for summing multiple metrics
from the daily table. Returns array<string, int>.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Comment thread src/Usage/Adapter/ClickHouse.php
Comment thread src/Usage/Adapter/ClickHouse.php Outdated
@lohanidamodar lohanidamodar changed the title Add ClickHouse adapter and Usage analytics library ClickHouse usage analytics: events/gauges tables with daily MV Apr 8, 2026
lohanidamodar and others added 4 commits April 8, 2026 08:00
Updated for events/gauges split, event-specific columns,
daily MV, query-time aggregation, billing methods, and
complete API reference with examples.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
…country/userAgent

- Remove stale 'type' key from addBatch() @param array shape in Usage.php, Adapter.php, Database.php
- Fix mixed-to-string cast in ClickHouse.php event column extraction with type-safe checks
- Reduce path size from 1024 to 255 and userAgent size from 512 to 255 in Metric::getEventSchema() to stay within MySQL 768-byte index limit
- Update MetricTest assertions: 11 attributes, 9 indexes, 7 EVENT_COLUMNS
- Update ClickHouseTest: userAgent/country are now event columns, not tags

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add UsageQuery class extending Query with a custom groupByInterval method
that enables time-bucketed aggregated queries. When present in the queries
array, the ClickHouse adapter switches from raw row returns to aggregated
results grouped by time bucket (SUM for events, argMax for gauges).

Supported intervals: 1m, 5m, 15m, 1h, 1d, 1w, 1M.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Comment thread src/Usage/Adapter/ClickHouse.php
Comment thread src/Usage/Usage.php
lohanidamodar and others added 2 commits April 9, 2026 05:45
Query::parse() uses static::isMethod() which allows UsageQuery
to extend the valid method list. Without this override, parsing
'groupByInterval("time","1h")' throws "Invalid query method".

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Add array shape type to findAggregatedFromTable $parsed parameter
  so PHPStan recognizes typed keys (filters, params, orderBy)
- Provide default for optional groupByInterval key access
- Split compound type check for $interval to satisfy PHPStan
  string narrowing in exception message interpolation
- Remove extra trailing blank line in UsageBase.php (PSR-12)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Comment thread src/Usage/Adapter/Database.php
Comment thread src/Usage/Adapter/Database.php
lohanidamodar and others added 2 commits April 16, 2026 04:51
The Database adapter silently cast string tenant IDs to int via (int),
which truncates non-numeric strings (e.g. UUIDs) to 0 — effectively
disabling tenant isolation. Now throws InvalidArgumentException for
non-numeric tenants so the mismatch is caught immediately.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
…on, getTotal ambiguity

- Buffer key now includes tag hash so events with same metric but
  different tags (e.g. different paths) stay as separate entries
  instead of silently discarding the second call's tags

- Daily table queries (findDaily, sumDaily, sumDailyBatch) now validate
  attributes against the daily schema (metric, value, time, tenant)
  instead of the full event schema. Querying path/method/status on the
  daily table now throws immediately instead of causing a ClickHouse
  "No such column" runtime error

- Changed (int) cast to (float) for agg_value in getTimeSeries to avoid
  truncating fractional gauge values or large event sums

- getTotal() now throws when a metric exists in both event and gauge
  tables instead of silently adding incompatible aggregations

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Comment on lines +2313 to +2323
if (is_array($json) && isset($json['data']) && is_array($json['data'])) {
foreach ($json['data'] as $row) {
$metricName = $row['metric'] ?? '';

if (!isset($totals[$metricName])) {
continue;
}

$totals[$metricName] += (int) ($row['agg_val'] ?? 0);
}
}
Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 getTotalBatch() silently sums semantically incompatible aggregations

When $type is null, the method queries both the events table (SUM(value)) and the gauges table (argMax(value, time)) and accumulates both results into $totals[$metricName] +=. If a metric name exists in both tables the result is the sum of a cumulative total and an instantaneous snapshot — a number with no valid meaning.

This is directly inconsistent with getTotal() (line 2137), which explicitly throws when a metric is found in both tables:

if ($eventTotal > 0 && $gaugeTotal > 0) {
    throw new Exception("Metric '{$metric}' exists in both event and gauge tables...");
}

Apply the same guard here, or at minimum document the known inconsistency.

Comment thread src/Usage/Adapter.php
Comment on lines +161 to +167
/**
* Enable or disable shared tables mode (multi-tenant with tenant column).
*
* @param bool $sharedTables
* @return self
*/
abstract public function setSharedTables(bool $sharedTables): self;
Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This seems specific to Database adapter, should we move it to Databases.php?

Comment thread src/Usage/Metric.php
Comment on lines +54 to +56
* - path: API endpoint path (events only)
* - method: HTTP method (events only)
* - status: HTTP status code (events only)
Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We're going to need a separate HttpLog.php type anyways, do we need these properties?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants