Files
scylladb/alternator/stats.cc
Nadav Har'El 70283967d3 alternator: add vector search metrics
Before this patch, we did not have any special metrics for vector search
in Alternator. We have had count of "Query" operations, but there was no
distinction between "standard" queries - of a base table or GSI/LSI -
and vector-search queries.

This patch adds four new metrics:

 * vector_search_query - counting how many Query requests are actually
   vector searches.

 * vector_search_query_returned_items - counting how many items were
   returned by vector searches.

 * vector_search_query_items_from_vs - counting how many results were
   retrieved from the vector-store backend.

 * vector_search_query_items_from_base_table - counting how many items
   were read from the base table during vector-search queries. Some
   vector search queries using SELECT=ALL_PROJECTED_ATTRIBUTES or COUNT
   are optimized to not need to read items from the base table.

This patch also includes documentation for the new four metrics, and
tests that they count what we want them to count.

Signed-off-by: Nadav Har'El <nyh@scylladb.com>
2026-05-13 11:57:44 +03:00

209 lines
17 KiB
C++

/*
* Copyright 2019-present ScyllaDB
*/
/*
* SPDX-License-Identifier: LicenseRef-ScyllaDB-Source-Available-1.1
*/
#include "stats.hh"
#include "utils/histogram_metrics_helper.hh"
#include <seastar/core/metrics.hh>
#include "utils/labels.hh"
namespace alternator {
const char* ALTERNATOR_METRICS = "alternator";
static seastar::metrics::label column_family_label("cf");
static seastar::metrics::label keyspace_label("ks");
static void register_metrics_with_optional_table(seastar::metrics::metric_groups& metrics, const stats& stats, const sstring& ks, const sstring& table) {
// Register the
seastar::metrics::label op("op");
bool has_table = table.length();
std::vector<seastar::metrics::label> aggregate_labels;
std::vector<seastar::metrics::label_instance> labels = {alternator_label};
sstring group_name = (has_table)? "alternator_table" : "alternator";
if (has_table) {
labels.push_back(column_family_label(table));
labels.push_back(keyspace_label(ks));
aggregate_labels.push_back(seastar::metrics::shard_label);
}
metrics.add_group(group_name, {
#define OPERATION(name, CamelCaseName) \
seastar::metrics::make_total_operations("operation", stats.api_operations.name, \
seastar::metrics::description("number of operations via Alternator API"), labels)(basic_level)(op(CamelCaseName)).aggregate(aggregate_labels).set_skip_when_empty(),
#define OPERATION_LATENCY(name, CamelCaseName) \
metrics.add_group(group_name, { \
seastar::metrics::make_histogram("op_latency", \
seastar::metrics::description("Latency histogram of an operation via Alternator API"), labels, [&stats]{return to_metrics_histogram(stats.api_operations.name.histogram());})(op(CamelCaseName))(basic_level).aggregate({seastar::metrics::shard_label}).set_skip_when_empty()}); \
if (!has_table) {\
metrics.add_group("alternator", { \
seastar::metrics::make_summary("op_latency_summary", \
seastar::metrics::description("Latency summary of an operation via Alternator API"), [&stats]{return to_metrics_summary(stats.api_operations.name.summary());})(op(CamelCaseName))(basic_level)(alternator_label).set_skip_when_empty()}); \
}
OPERATION(batch_get_item, "BatchGetItem")
OPERATION(batch_write_item, "BatchWriteItem")
OPERATION(create_backup, "CreateBackup")
OPERATION(create_global_table, "CreateGlobalTable")
OPERATION(delete_backup, "DeleteBackup")
OPERATION(delete_item, "DeleteItem")
OPERATION(describe_backup, "DescribeBackup")
OPERATION(describe_continuous_backups, "DescribeContinuousBackups")
OPERATION(describe_endpoints, "DescribeEndpoints")
OPERATION(describe_global_table, "DescribeGlobalTable")
OPERATION(describe_global_table_settings, "DescribeGlobalTableSettings")
OPERATION(describe_limits, "DescribeLimits")
OPERATION(describe_table, "DescribeTable")
OPERATION(describe_time_to_live, "DescribeTimeToLive")
OPERATION(get_item, "GetItem")
OPERATION(list_backups, "ListBackups")
OPERATION(list_global_tables, "ListGlobalTables")
OPERATION(list_tables, "ListTables")
OPERATION(list_tags_of_resource, "ListTagsOfResource")
OPERATION(put_item, "PutItem")
OPERATION(query, "Query")
OPERATION(restore_table_from_backup, "RestoreTableFromBackup")
OPERATION(restore_table_to_point_in_time, "RestoreTableToPointInTime")
OPERATION(scan, "Scan")
OPERATION(tag_resource, "TagResource")
OPERATION(transact_get_items, "TransactGetItems")
OPERATION(transact_write_items, "TransactWriteItems")
OPERATION(untag_resource, "UntagResource")
OPERATION(update_continuous_backups, "UpdateContinuousBackups")
OPERATION(update_global_table, "UpdateGlobalTable")
OPERATION(update_global_table_settings, "UpdateGlobalTableSettings")
OPERATION(update_item, "UpdateItem")
OPERATION(update_table, "UpdateTable")
OPERATION(update_time_to_live, "UpdateTimeToLive")
OPERATION(list_streams, "ListStreams")
OPERATION(describe_stream, "DescribeStream")
OPERATION(get_shard_iterator, "GetShardIterator")
OPERATION(get_records, "GetRecords")
});
OPERATION_LATENCY(put_item_latency, "PutItem")
OPERATION_LATENCY(get_item_latency, "GetItem")
OPERATION_LATENCY(delete_item_latency, "DeleteItem")
OPERATION_LATENCY(update_item_latency, "UpdateItem")
OPERATION_LATENCY(batch_write_item_latency, "BatchWriteItem")
OPERATION_LATENCY(batch_get_item_latency, "BatchGetItem")
OPERATION_LATENCY(get_records_latency, "GetRecords")
if (!has_table) {
// Create and delete operations are not applicable to a per-table metrics
// only register it for the global metrics
metrics.add_group("alternator", {
OPERATION(create_table, "CreateTable")
OPERATION(delete_table, "DeleteTable")
});
}
metrics.add_group(group_name, {
seastar::metrics::make_total_operations("unsupported_operations", stats.unsupported_operations,
seastar::metrics::description("number of unsupported operations via Alternator API"), labels).set_skip_when_empty(),
seastar::metrics::make_total_operations("total_operations", stats.total_operations,
seastar::metrics::description("number of total operations via Alternator API"), labels)(basic_level).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_total_operations("reads_before_write", stats.reads_before_write,
seastar::metrics::description("number of performed read-before-write operations"), labels).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_total_operations("write_using_lwt", stats.write_using_lwt,
seastar::metrics::description("number of writes that used LWT"), labels).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_total_operations("shard_bounce_for_lwt", stats.shard_bounce_for_lwt,
seastar::metrics::description("number writes that had to be bounced from this shard because of LWT requirements"), labels).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_total_operations("requests_blocked_memory", stats.requests_blocked_memory,
seastar::metrics::description("Counts a number of requests blocked due to memory pressure."), labels).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_total_operations("requests_shed", stats.requests_shed,
seastar::metrics::description("Counts a number of requests shed due to overload."), labels).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_total_operations("filtered_rows_read_total", stats.cql_stats.filtered_rows_read_total,
seastar::metrics::description("number of rows read during filtering operations"), labels).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_total_operations("filtered_rows_matched_total", stats.cql_stats.filtered_rows_matched_total,
seastar::metrics::description("number of rows read and matched during filtering operations"), labels).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_counter("rcu_total", [&stats]{return 0.5 * stats.rcu_half_units_total;},
seastar::metrics::description("total number of consumed read units"), labels).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_counter("wcu_total", stats.wcu_total[stats::wcu_types::PUT_ITEM],
seastar::metrics::description("total number of consumed write units"), labels)(op("PutItem")).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_counter("wcu_total", stats.wcu_total[stats::wcu_types::DELETE_ITEM],
seastar::metrics::description("total number of consumed write units"), labels)(op("DeleteItem")).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_counter("wcu_total", stats.wcu_total[stats::wcu_types::UPDATE_ITEM],
seastar::metrics::description("total number of consumed write units"), labels)(op("UpdateItem")).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_counter("wcu_total", stats.wcu_total[stats::wcu_types::INDEX],
seastar::metrics::description("total number of consumed write units"), labels)(op("Index")).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_total_operations("filtered_rows_dropped_total", [&stats] { return stats.cql_stats.filtered_rows_read_total - stats.cql_stats.filtered_rows_matched_total; },
seastar::metrics::description("number of rows read and dropped during filtering operations"), labels).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_counter("batch_item_count", seastar::metrics::description("The total number of items processed across all batches"), labels,
stats.api_operations.batch_write_item_batch_total)(op("BatchWriteItem")).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_counter("batch_item_count", seastar::metrics::description("The total number of items processed across all batches"), labels,
stats.api_operations.batch_get_item_batch_total)(op("BatchGetItem")).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_histogram("batch_item_count_histogram", seastar::metrics::description("Histogram of the number of items in a batch request"), labels,
[&stats]{ return to_metrics_histogram(stats.api_operations.batch_get_item_histogram);})(op("BatchGetItem")).aggregate({seastar::metrics::shard_label}).set_skip_when_empty(),
seastar::metrics::make_histogram("batch_item_count_histogram", seastar::metrics::description("Histogram of the number of items in a batch request"), labels,
[&stats]{ return to_metrics_histogram(stats.api_operations.batch_write_item_histogram);})(op("BatchWriteItem")).aggregate({seastar::metrics::shard_label}).set_skip_when_empty(),
seastar::metrics::make_histogram("operation_size_kb", seastar::metrics::description("Histogram of item sizes involved in a request"), labels,
[&stats]{ return to_metrics_histogram(stats.operation_sizes.get_item_op_size_kb);})(op("GetItem")).aggregate({seastar::metrics::shard_label}).set_skip_when_empty(),
seastar::metrics::make_histogram("operation_size_kb", seastar::metrics::description("Histogram of item sizes involved in a request"), labels,
[&stats]{ return to_metrics_histogram(stats.operation_sizes.put_item_op_size_kb);})(op("PutItem")).aggregate({seastar::metrics::shard_label}).set_skip_when_empty(),
seastar::metrics::make_histogram("operation_size_kb", seastar::metrics::description("Histogram of item sizes involved in a request"), labels,
[&stats]{ return to_metrics_histogram(stats.operation_sizes.delete_item_op_size_kb);})(op("DeleteItem")).aggregate({seastar::metrics::shard_label}).set_skip_when_empty(),
seastar::metrics::make_histogram("operation_size_kb", seastar::metrics::description("Histogram of item sizes involved in a request"), labels,
[&stats]{ return to_metrics_histogram(stats.operation_sizes.update_item_op_size_kb);})(op("UpdateItem")).aggregate({seastar::metrics::shard_label}).set_skip_when_empty(),
seastar::metrics::make_histogram("operation_size_kb", seastar::metrics::description("Histogram of item sizes involved in a request"), labels,
[&stats]{ return to_metrics_histogram(stats.operation_sizes.batch_get_item_op_size_kb);})(op("BatchGetItem")).aggregate({seastar::metrics::shard_label}).set_skip_when_empty(),
seastar::metrics::make_histogram("operation_size_kb", seastar::metrics::description("Histogram of item sizes involved in a request"), labels,
[&stats]{ return to_metrics_histogram(stats.operation_sizes.batch_write_item_op_size_kb);})(op("BatchWriteItem")).aggregate({seastar::metrics::shard_label}).set_skip_when_empty(),
});
seastar::metrics::label expression_label("expression");
metrics.add_group(group_name, {
seastar::metrics::make_total_operations("expression_cache_evictions", stats.expression_cache.evictions,
seastar::metrics::description("Counts number of entries evicted from expressions cache"), labels).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_total_operations("expression_cache_hits", stats.expression_cache.requests[stats::expression_types::UPDATE_EXPRESSION].hits,
seastar::metrics::description("Counts number of hits of cached expressions"), labels)(expression_label("UpdateExpression")).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_total_operations("expression_cache_misses", stats.expression_cache.requests[stats::expression_types::UPDATE_EXPRESSION].misses,
seastar::metrics::description("Counts number of misses of cached expressions"), labels)(expression_label("UpdateExpression")).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_total_operations("expression_cache_hits", stats.expression_cache.requests[stats::expression_types::CONDITION_EXPRESSION].hits,
seastar::metrics::description("Counts number of hits of cached expressions"), labels)(expression_label("ConditionExpression")).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_total_operations("expression_cache_misses", stats.expression_cache.requests[stats::expression_types::CONDITION_EXPRESSION].misses,
seastar::metrics::description("Counts number of misses of cached expressions"), labels)(expression_label("ConditionExpression")).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_total_operations("expression_cache_hits", stats.expression_cache.requests[stats::expression_types::PROJECTION_EXPRESSION].hits,
seastar::metrics::description("Counts number of hits of cached expressions"), labels)(expression_label("ProjectionExpression")).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_total_operations("expression_cache_misses", stats.expression_cache.requests[stats::expression_types::PROJECTION_EXPRESSION].misses,
seastar::metrics::description("Counts number of misses of cached expressions"), labels)(expression_label("ProjectionExpression")).aggregate(aggregate_labels).set_skip_when_empty()
});
// Vector search metrics
metrics.add_group(group_name, {
seastar::metrics::make_total_operations("vector_search_query", stats.vector_search.query,
seastar::metrics::description("number of Query operations with VectorSearch"), labels).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_total_operations("vector_search_query_returned_items", stats.vector_search.query_returned_items,
seastar::metrics::description("total number of items returned by Query operations with VectorSearch"), labels).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_total_operations("vector_search_query_items_from_vs", stats.vector_search.query_items_from_vs,
seastar::metrics::description("total number of nearest neighbors found by the vector store (some may be post-filtered and not returned)"), labels).aggregate(aggregate_labels).set_skip_when_empty(),
seastar::metrics::make_total_operations("vector_search_query_items_from_base_table", stats.vector_search.query_items_from_base_table,
seastar::metrics::description("total number of items read from the base table by vector search queries"), labels).aggregate(aggregate_labels).set_skip_when_empty(),
});
// Only register the following metrics for the global metrics, not per-table
if (!has_table) {
metrics.add_group("alternator", {
seastar::metrics::make_counter("authentication_failures", stats.authentication_failures,
seastar::metrics::description("total number of authentication failures"), labels).aggregate({seastar::metrics::shard_label}).set_skip_when_empty(),
seastar::metrics::make_counter("authorization_failures", stats.authorization_failures,
seastar::metrics::description("total number of authorization failures"), labels).aggregate({seastar::metrics::shard_label}).set_skip_when_empty(),
});
}
}
void register_metrics(seastar::metrics::metric_groups& metrics, const stats& stats) {
register_metrics_with_optional_table(metrics, stats, "", "");
}
table_stats::table_stats(const sstring& ks, const sstring& table) {
_stats = make_lw_shared<stats>();
register_metrics_with_optional_table(_metrics, *_stats, ks, table);
}
}