DSE Performance Service diagnostic table reference

Reference information on performance diagnostic information in a set of data dictionary tables can be queried with CQL. The following types of performance service diagnostic tables are available:

CQL slow log table
CQL system info tables
Data Resource latency tracking tables
Database summary statistics tables
Cluster summary statistics tables
Histogram tables
User and resource latency tracking tables

Table names that contain _snapshot are not related to nodetool snapshot. These tables are snapshots of the data in the last few seconds of activity in the system.

CQL slow log table

`node_slow_log` table

Queries on a node exceeding the `threshold_ms` parameter.
Column Name	Data type	Description
`node_ip`	inet	Node address.
date	timestamp	Date of entry (MM/DD/YYYY granularity).
`start_time`	timeuuid	Start timestamp of query execution.
commands	list_text_	CQL statements being executed.
duration	bigint	Execution time in milliseconds.
parameters	map_text_	Not used at this time.
`source_ip`	inet	Client address.
`table_names`	set_text_	CQL tables touched.
username	text	User executing query, if authentication is enabled.

CQL system info tables

`key_cache` table

Key cache performance statistics.
Column Name	Data type	Description
`node_ip`	inet	Node address.
`cache_capacity`	bigint	Key cache capacity in bytes.
`cache_hits`	bigint	Total number of cache hits since startup.
`cache_requests`	bigint	Total number of cache requests since startup.
`cache_size`	bigint	Current key cache size in bytes.
`hit_rate`	double	Ratio of hits to requests since startup.

`net_stats` table

Data flow operations repair tasks and more.
Column Name	Data type	Description
`node_ip`	inet	Node address.
`read_repair_attempted`	bigint	Read repairs attempted since startup.
`read_repaired_blocking`	bigint	Number of read repairs performed syncronously since startup.
`read_repaired_background`	bigint	Number of read repairs performed asyncronously since startup.
`commands_pending`	int	Current number of read repair commands pending.
`commands_completed`	bigint	Total read repair commands completed since startup.
`responses_pending`	int	Current read repair responses pending count.
`responses_completed`	bigint	Current read repairs completed count.

`thread_pool` table

Information on thread pool activity.
Column Name	Data type	Description
`node_ip`	inet	Node address.
`pool_name`	text	Thread pool name.
active	bigint	Currently active tasks.
`all_time_blocked`	bigint	Total blocked tasks since startup.
blocked	bigint	Currently blocked tasks.
completed	bigint	Total completed tasks since startup.
pending	bigint	Currently pending tasks.

`thread_pool_messages` table

Information about thread pool messages
Column Name	Data type	Description
`node_ip`	inet	Node address.
`message_type`	text	Inter-node message type.
`dropped_count`	int	Total count of dropped messages since startup.

Data Resource latency tracking tables

`object_io` table

Per node recent latency metrics by keyspace and table.
Column Name	Data type	Description
`node_ip`	inet	Node address.
`keyspace_name`	text	Keyspace name.
`table_name`	text	Table name.
`last_activity`	timestamp	End of sampling period in which this object was last active.
`memory_only`	boolean	DSE memory only table.
`read_latency`	double	Mean value in microseconds for all reads during the last active sampling period for this object.
`total_reads`	bigint	Count during the last active sampling period for this object.
`total_writes`	bigint	Count during the last active sampling period for this object.
`write_latency`	double	Mean value in microseconds for all writes during the last active sampling period for this object.

`user_object_read_io_snapshot` table

`user_object_read_io_snapshot` table ordered by user/object and user total write latency
Column Name	Data type	Description
`keyspace_name`	text	Keyspace name.
`table_name`	text	Table name.
`node_ip`	inet	Node address.
`user_ip`	inet	User node address.
`conn_ip`	inet	Connection node address.
username	text	User name.
`read_latency`	double	In microseconds during the last sampling period.
`total_reads`	bigint	Count during the last sampling period.
`write_latency`	double	In microseconds during the last sampling period.
`total_writes`	bigint	Count during the last sampling period.
`latency_index`	int	Ranking by mean read latency during the last sampling period.
`read_quantiles`	boolean	DSE memory only table.

`object_read_io_snapshot` table

Per node recent latency metrics by keyspace and table. Ordered by user/object and user total write latency.
Column Name	Data type	Description
`keyspace_name`	text	Keyspace name.
`table_name`	text	Table name.
`node_ip`	inet	Node address.
`user_ip`	inet	User node address.
`conn_ip`	inet	Connection node address.
username	text	User name.
`read_latency`	double	In microseconds during the last sampling period.
`total_reads`	bigint	Count during the last sampling period.
`write_latency`	double	In microseconds during the last sampling period.
`total_writes`	bigint	Count during the last sampling period.
`latency_index`	int	Ranking by mean read latency during the last sampling period.
`read_quantiles`	boolean	DSE memory only table.

`object_write_io_snapshot` table

Per node recent latency metrics by keyspace and table. Ordered by mean write latency. Scale of 0 to 99 (0 is worst)
Column Name	Data type	Description
`node_ip`	inet	Node address.
`latency_index`	int	Ranking by mean write latency during the last sampling period.
`keyspace_name`	text	Keyspace name.
`memory_only`	boolean	DSE memory only table.
`read_latency`	double	Mean value in microseconds during the active sampling period.
`table_name`	text	Table name.
`total_reads`	bigint	Count during the last sampling period.
`total_writes`	bigint	Count during the last sampling period.
`write_latency`	double	Mean value in microseconds during the last sampling period.

Database summary statistics tables

`node_table_snapshot` table

Per node table metrics by keyspace and table
Column Name	Data type	Description
`node_ip`	inet	Node address.
`keyspace_name`	text	Keyspace name.
`table_name`	text	Table name.
`bf_false_positive_ratio`	double	Bloom filter false positive ratio since startup.
`bf_false_positives`	bigint	Bloom filter false positive count since startup.
`compression_ratio`	double	Current compression ratio of SSTables.
`droppable_tombstone_ratio`	double	Ratio of tombstones older than `gc_grace_seconds` against total column count in all SSTables.
`key_cache_hit_rate`	double	Current key cache hit rate.
`live_sstable_count`	bigint	Current SSTable count.
`max_row_size`	bigint	Maximum partition size in bytes.
`mean_read_latency`	double	In microseconds for this table since startup.
`mean_row_size`	bigint	Average partition size in bytes.
`mean_write_latency`	double	In microseconds for this table since startup.
`memtable_columns_count`	bigint	Approximate number of cells for this table currently resident in memtables.
`memtable_size`	bigint	Total size in bytes of memtable data.
`memtable_switch_count`	bigint	Number of times memtables have been flushed since startup.
`min_row_size`	bigint	Minimum partition size in bytes.
`total_data_size`	bigint	Data size on disk in bytes.
`total_reads`	bigint	Number of reads since startup.
`total_writes`	bigint	Number of writes since startup.
`unleveled_sstables`	bigint	Current count of SSTables in level 0 (if using leveled compaction).

`table_snapshot` table

Cluster-wide lifetime table metrics by keyspace and table. This table aggregates `node_table_snapshot` from each node in the cluster.
Column Name	Data type	Description
`keyspace_name`	text	Keyspace name.
`table_name`	text	Table name.
`bf_false_positive_ratio`	double	Bloom filter false positive ratio since startup.
`bf_false_positives`	bigint	Bloom filter false positive count since startup.
`compression_ratio`	double	Current compression ratio of SSTables.
`droppable_tombstone_ratio`	double	Ratio of tombstones older than `gc_grace_seconds` against total column count in all SSTables.
`key_cache_hit_rate`	double	Current key cache hit rate.
`live_sstable_count`	bigint	Current SSTable count.
`max_row_size`	bigint	Maximum partition size in bytes.
`mean_read_latency`	double	In microseconds for this table since startup.
`mean_row_size`	bigint	Average partition size in bytes.
`mean_write_latency`	double	In microseconds for this table since startup.
`memtable_columns_count`	bigint	Approximate number of cells for this table currently resident in memtables.
`memtable_size`	bigint	Total size in bytes of memtable data.
`memtable_switch_count`	bigint	Number of times memtables have been flushed since startup.
`min_row_size`	bigint	Minimum partition size in bytes.
`total_data_size`	bigint	Data size on disk in bytes.
`total_reads`	bigint	Number of reads since startup.
`total_writes`	bigint	Number of writes since startup.
`unleveled_sstables`	bigint	Current count of SSTables in level 0 (if using leveled compaction).

`keyspace_snapshot` table

Cluster wide lifetime table metrics, aggregated at the keyspace level (aggregates the data in `table_snapshot`).
Column Name	Data type	Description
`keyspace_name`	text	Keyspace name.
`index_count`	int	Number of secondary indexes.
`mean_read_latency`	double	For all tables in the keyspace and all nodes in the cluster since startup.
`mean_write_latency`	double	For all tables in the keyspace and all nodes in the cluster since startup.
`table_count`	int	Number of tables in the keyspace.
`total_data_size`	bigint	Total size in bytes of SSTables for all tables and indexes across all nodes in the cluster.
`total_reads`	bigint	For all tables, across all nodes.
`total_writes`	bigint	For all tables, across all nodes.

Cluster summary statistics tables

`node_snapshot` table

Per node system metrics.
Column Name	Data type	Description
`node_ip`	inet	Node address.
`cms_collection_count`	bigint	CMS garbage collections since startup.
`cms_collection_time`	bigint	Total time spent in CMS garbage collection since startup.
`commitlog_pending_tasks`	bigint	Current commit log tasks pending.
`commitlog_size`	bigint	Total commit log size in bytes.
`compactions_completed`	bigint	Number of compactions completed since startup.
`compactions_pending`	int	Number of pending compactions.
`completed_mutations`	bigint	Total number of mutations performed since startup.
`data_owned`	float	Percentage of total data owned by this node.
datacenter	text	Datacenter name.
`dropped_mutation_ratio`	double	Ratio of dropped to completed mutations since startup.
`dropped_mutations`	bigint	Total number of dropped mutations since startup.
`flush_sorter_tasks_pending`	bigint	Current number of memtable flush sort tasks pending.
`free_space`	bigint	Total free disk space in bytes.
`gossip_tasks_pending`	bigint	Current number of gossip tasks pending.
`heap_total`	bigint	Total available heap memory in bytes.
`heap_used`	bigint	Current heap usage in bytes.
`hinted_handoff_pending`	bigint	Current number of hinted handoff tasks pending.
`index_data_size`	bigint	Total size in bytes of index column families.
`internal_responses_pending`	bigint	Current number of internal response tasks pending.
`key_cache_capacity`	bigint	Key cache capacity in bytes.
`key_cache_entries`	bigint	Current number of key cache entries.
`key_cache_size`	bigint	Current key cache size in bytes.
`manual_repair_tasks_pending`	bigint	Current number of manual repair tasks pending.
`mean_range_slice_latency`	double	Mean latency in microseconds for range slice operations since startup.
`mean_read_latency`	double	Mean latency in microseconds for reads since startup.
`mean_write_latency`	double	Mean latency in microseconds for writes since startup.
`memtable_post_flushers_pending`	bigint	Current number of memtable post flush tasks pending.
`migrations_pending`	bigint	Current number of migration tasks pending.
`misc_tasks_pending`	bigint	Current number of misc tasks pending.
`parnew_collection_count`	bigint	ParNew garbage collections since startup.
`parnew_collection_time`	bigint	Total time spent in ParNew garbage collection since startup.
`process_cpu_load`	double	Current CPU load for the DSE process (Linux only).
rack	text	Rack identifier.
`range_slice_timeouts`	bigint	Number of timed out range slice requests since startup.
`read_repair_tasks_pending`	bigint	Current number of read repair tasks pending.
`read_requests_pending`	bigint	Current read requests pending.
`read_timeouts`	bigint	Number of timed out range slice requests since startup.
`replicate_on_write_tasks_pending`	bigint	Current.
`request_responses_pending`	bigint	Current.
`row_cache_capacity`	bigint	Row cache capacity in bytes
`row_cache_entries`	bigint	Current number of row cache entries.
`row_cache_size`	bigint	Current row cache size in bytes.
state	text	Node State (`JOINING`/`LEAVING`/`MOVING`/`NORMAL`).
`storage_capacity`	bigint	Total disk space in bytes.
`streams_pending`	int	Current number of pending streams.
`table_data_size`	bigint	Total size in bytes of non-index column families.
tokens	settext	Tokens owned by the this node.
`total_batches_replayed`	bigint	Total number of batchlog entries replayed since startup
`total_node_memory`	bigint	Total available RAM (Linux only).
`total_range_slices`	bigint	Total number of range slice operations performed since startup.
`total_reads`	bigint	Total number of reads performed since startup.
`total_writes`	bigint	Total number of writes performed since startup.
uptime	bigint	Node uptime in seconds.
`write_requests_pending`	bigint	Total number of write tasks pending.
`write_timeouts`	bigint	Number of timed out range slice requests since startup.

`dc_snapshot` table

Aggregates `node_snapshot` data at the datacenter level.
Column Name	Data type	Description
name	text	Datacenter name
`compactions_completed`	bigint	Total number of compactions completed since startup by all nodes in the data center.
`compactions_pending`	int	Total number of pending compactions on all nodes in the datacenter.
`completed_mutations`	bigint	Total number of mutations performed since startup by all nodes in the data center.
`dropped_mutation_ratio`	double	Ratio of dropped to completed mutations since startup across all nodes in the datacenter.
`dropped_mutations`	bigint	Total number of dropped mutations since startup by all nodes in the data center.
`flush_sorter_tasks_pending`	bigint	Total number of memtable flush sort tasks pending across all nodes in the datacenter.
`free_space`	bigint	Total free disk space in bytes across all nodes in the datacenter.
`gossip_tasks_pending`	bigint	Total number of gossip tasks pending across all nodes in the data center.
`hinted_handoff_pending`	bigint	Total number of hinted handoff tasks pending across all nodes in the data center.
`index_data_size`	bigint	Total size in bytes of index column families across all nodes in the data center.
`internal_responses_pending`	bigint	number of internal response tasks pending across all nodes in the data center.
`key_cache_capacity`	bigint	Total capacity in bytes of key caches across all nodes in the data center.
`key_cache_entries`	bigint	Total number of entries in key caches across all nodes in the data center.
`key_cache_size`	bigint	Total consumed size in bytes of key caches across all nodes in the data center.
`manual_repair_tasks_pending`	bigint	Total number of manual repair tasks pending across all nodes in the data center.
`mean_range_slice_latency`	double	Mean latency in microseconds for range slice operations, averaged across all nodes in the datacenter.
`mean_read_latency`	double	Mean latency in microseconds for read operations, averaged across all nodes in the datacenter.
`mean_write_latency`	double	Mean latency in microseconds for write operations, averaged across all nodes in the datacenter.
`memtable_post_flushers_pending`	bigint	Total number of memtable post flush tasks pending across all nodes in the datacenter.
`migrations_pending`	bigint	Total number of migration tasks pending across all nodes in the data center.
`misc_tasks_pending`	bigint	Total number of misc tasks pending across all nodes in the datacenter.
`node_count`	int	Total number of live nodes in the datacenter.
`read_repair_tasks_pending`	bigint	Total number of read repair tasks pending across all nodes in the data center.
`read_requests_pending`	bigint	Total read requests pending across all nodes in the datacenter.
`replicate_on_write_tasks_pending`	bigint	Total number of counter replicate on write tasks pending across all nodes in the datacenter.
`request_responses_pending`	bigint	Total number of request response tasks pending across all nodes in the data center.
`row_cache_capacity`	bigint	Total capacity in bytes of partition caches across all nodes in the data center.
`row_cache_entries`	bigint	Total number of row cache entries all nodes in the datacenter.
`row_cache_size`	bigint	Total consumed size in bytes of row caches across all nodes in the data center.
`storage_capacity`	bigint	Total disk space in bytes across all nodes in the datacenter.
`streams_pending`	int	Number of pending streams across all nodes in the datacenter.
`table_data_size`	bigint	Total size in bytes of non-index column families across all nodes in the data center.
`total_batches_replayed`	bigint	Total number of batchlog entries replayed since startup by all nodes in the datacenter.
`total_range_slices`	bigint	Total number of range slice operations performed since startup by all nodes in the datacenter.
`total_reads`	bigint	Total number of read operations performed since startup by all nodes in the datacenter.
`total_writes`	bigint	Total number of write operations performed since startup by all nodes in the datacenter.
`write_requests_pending`	bigint	Total number of write tasks pending across all nodes in the data center.

`cluster_snapshot` table

`cluster_snapshot` aggregates `node_shapshot` data for the entire cluster.
Column Name	Data type	Description
name	text	Cluster name.
`compactions_completed`	bigint	Total number of compactions completed since startup by all nodes in the cluster.
`completed_mutations`	bigint	Total number of mutations performed since startup by all nodes in the cluster.
`compactions_pending`	int	Total number of pending compactions on all nodes in the cluster.
datacenters	settext	Datacenter names.
`dropped_mutation_ratio`	double	Ratio of dropped to completed mutations since startup across all nodes in the cluster.
`dropped_mutations`	bigint	Total number of dropped mutations since startup by all nodes in the cluster.
`flush_sorter_tasks_pending`	bigint	Total number of memtable flush sort tasks pending across all nodes in the cluster.
`free_space`	bigint	Total free disk space in bytes across all nodes in the cluster.
`gossip_tasks_pending`	bigint	Total number of gossip tasks pending across all nodes in the cluster.
`hinted_handoff_pending`	bigint	Total number of hinted handoff tasks pending across all nodes in the cluster.
`index_data_size`	bigint	Total size in bytes of index column families across all nodes in the cluster.
`internal_responses_pending`	bigint	Number of internal response tasks pending across all nodes in the cluster.
`key_cache_capacity`	bigint	Total capacity in bytes of key caches across all nodes in the cluster.
`key_cache_entries`	bigint	Total number of entries in key caches across all nodes in the cluster.
`key_cache_size`	bigint	Total consumed size in bytes of key caches across all nodes in the cluster.
`keyspace_count`	int	Total number of keyspaces defined in schema.
`manual_repair_tasks_pending`	bigint	Total number of manual repair tasks pending across all nodes in the cluster.
`mean_range_slice_latency`	double	Mean latency in microseconds for range slice operations, averaged across all nodes in the cluster.
`mean_read_latency`	double	Mean latency in microseconds for read operations, averaged across all nodes in the cluster.
`mean_write_latency`	double	Mean latency in microseconds for write operations, averaged across all nodes in the cluster.
`memtable_post_flushers_pending`	bigint	Total number of memtable post flush tasks pending across all nodes in the cluster.
`migrations_pending`	bigint	Total number of migration tasks pending across all nodes in the cluster.
`misc_tasks_pending`	bigint	Total number of misc tasks pending across all nodes in the cluster.
`node_count`	int	Total number of live nodes in the cluster.
`read_repair_tasks_pending`	bigint	Total number of read repair tasks pending across all nodes in the cluster.
`read_requests_pending`	bigint	Total read requests pending across all nodes in the cluster.
`replicate_on_write_tasks_pending`	bigint	Total number of counter replicate on write tasks pending across all nodes in the cluster.
`request_responses_pending`	bigint	Total number of request response tasks pending across all nodes in the cluster
`row_cache_capacity`	bigint	Total capacity in bytes of partition caches across all nodes in the cluster.
`row_cache_entries`	bigint	Total number of row cache entries all nodes in the cluster.
`row_cache_size`	bigint	Total consumed size in bytes of row caches across all nodes in the cluster
`storage_capacity`	bigint	Total disk space in bytes across all nodes in the cluster.
`streams_pending`	int	Number of pending streams across all nodes in the cluster.
`table_count`	int	Total number of tables defined in schema.
`table_data_size`	bigint	Total size in bytes of non-index column families across all nodes in the cluster.
`total_batches_replayed`	bigint	Total number of batchlog entries replayed since startup by all nodes in the cluster.
`total_range_slices`	bigint	Total number of read operations performed since startup by all nodes in the cluster.
`total_reads`	bigint	Total number of write operations performed since startup by all nodes in the cluster.
`total_writes`	bigint	Total number of write tasks pending across all nodes in the cluster.
`write_requests_pending`	bigint	Total number of write tasks pending across all nodes in the cluster.

A histogram measures the distribution of values in a stream of data. These histogram tables use the Metrics Core library. You must enable the collection of table histogram data using the DataStax Enterprise Performance Service.

These tables show similar information to the data obtained by the nodetool tablehistograms utility. The major difference is that the nodetool histograms output is recent data, while the diagnostic tables contain lifetime data. The data in the diagnostic histogram tables is cumulative since the DSE server was started. In contrast, the nodetool tablehistograms shows the values for the past fifteen minutes.

Histogram tables provide DSE statistics that can be queried with CQL and are generated with these templates:

Detailed
Summary
Global

DSE provided histogram tables
Histogram	Details table	Summary table	Keyspace details	Keyspace summary	Global details	Global summary
`cell_count`	Y	Y	N	N	N	N
`partition_size`	Y	Y	N	N	N	N
`range_latency`	Y	Y	Y	N	Y	N
`read_latency`	Y	Y	Y	N	Y	N
`sstables_per_read`	Y	Y	Y	N	N	N
`write_latency`	Y	Y	Y	N	N	N

Summary histogram tables

Summary histogram tables the percentages with output similar to the nodetool tablehistograms command output.

Summary histogram tables
Column Name	Data type	Description
`node_ip`	inet	Node address
`keyspace_name`	text	Keyspace name
`table_name`	text	Table name
`histogram_id`	timestamp	Groups rows by the specific histogram they belong to. Rows for the same node, keyspace and table are ordered by this field, to enable date-based filtering.
p50	bigint	The threshold where 50 percent of the operation is recorded 50% from the end, for the 50th percentile.
p75	bigint	The threshold where 75 percent of the operation is recorded 25% from the end, for the 75th percentile.
p90	bigint	The threshold where 90 percent of the operation is recorded 10% from the end, for the 90th percentile.
p95	bigint	The threshold where 95 percent of the operation is recorded 5% from the end, for the 95th percentile.
p98	bigint	The threshold where 98 percent of the operation is recorded 2% from the end, for the 98th percentile.
p99	bigint	The threshold where 99 percent of the operation is recorded 1% from the end, for the 99th percentile.
min	bigint	The minimum number of operations.
max	bigint	The maximum number of operations.
`dropped_messages`	bigint	The total number of dropped messages for mutations to this table.

Detailed histogram for keyspaces

Detailed data for a single keyspace.

Detailed histogram for keyspaces
Column Name	Data type	Description
`node_ip`	inet	Node address
`keyspace_name`	text	Keyspace name
`table_name`	text	Table name
`histogram_id`	timestamp	Groups rows by the specific histogram they belong to. Rows for the same node, keyspace and table are ordered by this field, to enable date-based filtering.
`bucket_offset`	bigint	The number between the current bucket and the previous bucket.
`bucket_count`	bigint	The sum of values being measured that is less than or equal to this offset and greater than or equal to the previous offset.

Detailed table histogram

Detailed data for a single table.

Detailed table histogram
Column Name	Data type	Description
`node_ip`	inet	Node address
`keyspace_name`	text	Keyspace name
`table_name`	text	Table name
`histogram_id`	timestamp	Groups rows by the specific histogram they belong to. Rows for the same node, keyspace and table are ordered by this field, to enable date-based filtering.
`bucket_offset`	bigint	The number between the current bucket and the previous bucket.
`bucket_count`	bigint	The sum of values being measured that is less than or equal to this offset and greater than or equal to the previous offset.

Global histogram tables

Detailed data for all tables and all keyspaces, cumulative since the last time the node was started.

Global histogram tables
Column Name	Data type	Description
`node_ip`	inet	Node address.
`keyspace_name`	text	Keyspace name.
`table_name`	text	Table name.
`histogram_id`	timestamp	Groups rows by the specific histogram they belong to. Rows for the same node, keyspace, and table are ordered by this field, to enable date-based filtering.
`bucket_offset`	bigint	Number of cells in a partition.
`bucket_count`	bigint	Number of partitions where the cell count falls in the corresponding bucket.

`cell_count_histograms` table

Cell count per partition histogram data.
Column Name	Data type	Description
`node_ip`	inet	Node address.
`histogram_id`	timestamp	Groups rows by the specific histogram they belong to. Rows for the same node, keyspace and table are ordered by this field, to enable date-based filtering.
`bucket_offset`	bigint	The number between the current bucket and the previous bucket.
`bucket_count`	bigint	The sum of values being measured that is less than or equal to this offset and greater than or equal to the previous offset.

`dropped_messages` table

Dropped messages histogram data in seconds
Column Name	Data type	Description
`node_ip`	inet	Node address.
`histogram_id`	timestamp	The timestamp when the histogram record was written.
verb	text	Where verb denotes the message type that was dropped: `MUTATION`, `HINT`, `READ_REPAIR`, `READ`, `REQUEST_RESPONSE`, `BATCH_STORE`, `BATCH_REMOVE`, `RANGE_SLICE`, `GOSSIP_DIGEST_SYN`, `GOSSIP_DIGEST_ACK`, `GOSSIP_DIGEST_ACK2`, `DEFINITIONS_UPDATE`, `TRUNCATE`, `SCHEMA_CHECK`, `REPLICATION_FINISHED`, `INTERNAL_RESPONSE`, `COUNTER_MUTATION`, `SNAPSHOT`, `MIGRATION_REQUEST`, `GOSSIP_SHUTDOWN`, `ECHO`, `REPAIR_MESSAGE`, `PAXOS_PREPARE`, `PAXOS_PROPOSE`, `PAXOS_COMMIT`.
`global_count`	bigint	Global metrics are the sum of the internal and the cross-node metrics for dropped events since the server was started, including dropped mutations.
`global_mean_rate`	double	Global metrics for dropped messages, including dropped mutations.
`global_1min_rate`	double	Estimated rate of the combined internal and the cross-node metrics for dropped messages for 1 minute.
`global_5min_rate`	double	Estimated sum of the internal and the cross-node metrics for dropped messages
`global_15min_rate`	double	Global metrics for dropped messages for 15 minutes.
`table_name`	text	Table name.
`internal_count`	bigint	Inside one DSE node, the number of internal messages that were dropped since the server was started.
`internal_mean_rate`	double	Inside one DSE node, the average rate of dropped message events per second.
`internal_1min_rate`	double	Inside one DSE node, the average number of messages that were dropped in a one-minute interval.
`internal_5min_rate`	double	Inside one DSE node, the average number of messages that were dropped in a five-minute interval.
`internal_15min_rate`	double	Inside one DSE node, the average number of messages that were dropped in a 15-minute interval.
`internal_latency_median`	double	Inside one DSE node, the median of all recorded durations for one second.
`internal_latency_p75`	double	Inside one DSE node, the threshold where 75 percent of the latency is recorded 25% from the end, for the 75th percentile.
`internal_latency_p90`	double	Inside one DSE node, the threshold where 90 percent of the latency is recorded 10% from the end, for the 90th percentile.
`internal_latency_p95`	double	Inside one DSE node, the threshold where 95 percent of the latency is recorded 5% from the end, for the 95th percentile.
`internal_latency_p98`	double	Inside one DSE node, the threshold where 98 percent of the latency is recorded 2% from the end, for the 98th percentile.
`internal_latency_p99`	double	Inside one DSE node, the threshold where 99 percent of the latency is recorded 1% from the end, for the 99th percentile.
`internal_latency_min`	double	Inside one DSE node, the minimum number of dropped mutations.
`internal_latency_mean`	double	Inside one DSE node, the average number of messages dropped.
`internal_latency_max`	double	Inside one DSE node, the maximum number of dropped mutations.
`internal_latency_stdev`	double	Inside one DSE node, the standard deviation of dropped mutations.
`xnode_count`	bigint	For cross node messages, the number of messages that were dropped since the server was started.
`xnode_mean_rate`	double	For cross node messages, the average number of messages dropped.
`xnode_1min_rate`	double	For cross node messages, the average number of messages that were dropped in a one-minute interval.
`xnode_5min_rate`	double	For cross node messages, the average number of messages that were dropped in a five-minute interval.
`xnode_15min_rate`	double	For cross node messages, the average number of messages that were dropped in a 15-minute interval.
`xnode_median`	double	For cross node messages, the median of all recorded durations for one second.
`xnode_p75`	double	For cross node messages, the threshold where 75 percent of the latency is recorded 25% from the end, for the 75th percentile.
`xnode_p90`	double	For cross node messages, the threshold where 90 percent of the latency is recorded 10% from the end, for the 90th percentile.
`xnode_p95`	double	For cross node messages, the threshold where 95 percent of the latency is recorded 5% from the end, for the 95th percentile.
`xnode_p98`	double	For cross node messages, the threshold where 98 percent of the latency is recorded 2% from the end, for the 98th percentile.
`xnode_p99`	double	For cross node messages, the maximum number dropped messages are recorded 1% from the end, for the 99th percentile.
`xnode_min`	double	For cross node messages, the minmum number of dropped messages per second.
`xnode_mean`	double	For cross node messages, the average number of dropped messages per second.
`xnode_max`	double	For cross node messages, the maximum number of dropped messages per second.
`xnode_stdev`	double	For cross node messages, the standard deviation for dropped messages per second.
`bucket_offset`	bigint	The number between the current bucket and the previous bucket.
`bucket_count`	bigint	The sum of values being measured that is less than or equal to this offset and greater than or equal to the previous offset.

User and resource latency tracking tables

`user_io` table

`user_io` table per node, long-lived read/write metrics by client connection and aggregated for all keyspaces and tables.
Column Name	Data type	Description
`node_ip`	inet	Node address.
`conn_id`	text	Unique client connection ID.
`last_activity`	timestamp	End of sampling period in which this client was last active.
`read_latency`	double	In microseconds for the last active sampling period.
`total_reads`	bigint	Count during the last active sampling period for this client.
`total_writes`	bigint	Count during the last active sampling period for this client.
`user_ip`	inet	Client origin address.
username	text	Present if authentication is enabled.
`write_latency`	double	In microseconds for the last active sampling period.

`user_read_io_snapshot` table

`user_read_io_snapshot` per node recent read/write metrics by keyspace, table, and client during the most recent sampling period.
Column Name	Data type	Description
`node_ip`	inet	Node address.
`latency_index`	int	Ranking by mean read latency during the last sampling period.
`conn_id`	text	Unique client connection ID.
`read_latency`	double	Mean value in microseconds during the last sampling period.
`total_reads`	bigint	During the last sampling period.
`total_writes`	bigint	During the last sampling period.
`user_ip`	inet	Client origin address.
username	text	Present if authentication is enabled.
`write_latency`	double	Mean value in microseconds during the last sampling period.

`user_write_io_snapshot` table

`user_write_io_snapshot` per node recent read/write metrics by keyspace, table, and client during the most recent sampling period.
Column Name	Data type	Description
`node_ip`	inet	Node address.
`latency_index`	int	Ranking by mean write latency during the last sampling period.
`conn_id`	text	Unique client connection ID.
`read_latency`	double	Mean value in microseconds during the last sampling period.
`total_reads`	bigint	During the last sampling period.
`total_writes`	bigint	During the last sampling period.
`user_ip`	inet	Client origin address.
username	text	Present if authentication is enabled.
`write_latency`	double	Mean value in microseconds during the last sampling period.

`user_object_io` table

Per node, long-lived read/write metrics by client connection, keyspace and table.
Column Name	Data type	Description
`node_ip`	inet	Node address.
`conn_id`	text	Unique client connection ID.
`keyspace_name`	text	Keyspace name.
`table_name`	text	Table name.
`last_activity`	timestamp	End of sampling period in which this client was last active against this object.
`read_latency`	double	Mean value in microseconds during the last active sampling period for this object/client.
`total_reads`	bigint	During the last active sampling period for this object/client.
`total_writes`	bigint	During the last active sampling period for this object/client.
`user_ip`	inet	Client origin address.
username	text	Present if authentication is enabled.
`write_latency`	double	Mean value in microseconds during the last active sampling period for this object/client.

`user_object_write_io_snapshot` table

Per node recent read/write metrics by client, keyspace, and table during the most recent sampling period.
Column Name	Data type	Description
`node_ip`	inet	Node address.
`latency_index`	int	Ranking by mean write latency during the last sampling period.
`conn_id`	text	Unique client connection ID.
`keyspace_name`	text	Keyspace name.
`read_latency`	double	Mean value in microseconds during the last sampling period.
`table_name`	text	Table name.
`total_reads`	bigint	During the last sampling period.
`total_writes`	bigint	During the last sampling period.
`user_ip`	inet	Client origin address.
username	text	Present if authentication is enabled.
`write_latency`	double	Mean value in microseconds during the last sampling period.

`object_user_io` table

Overview of the I/O activity by user for each table.
Column Name	Data type	Description
`node_ip`	inet	Node address.
`keyspace_name`	text	Keyspace name.
`table_name`	text	Table name.
`conn_id`	text	Unique client connection ID.
`last_activity`	timestamp	End of sampling period in which this client connection was last active against this object.
`read_latency`	double	Mean value in microseconds during the last active sampling period for this object/client.
`total_reads`	bigint	Count during the last active sampling period for this object/client.
`total_writes`	bigint	Count during the last active sampling period for this object/client.
`user_ip`	inet	Client origin address.
username	text	Present if authentication is enabled.
`write_latency`	double	Mean value in microseconds during the last active sampling period for this object/client.

`object_user_read_io_snapshot` table

Per node recent read/write metrics by client, keyspace, and table during the most recent sampling period. Tracks best-worst latency on a scale of 0 (worst) to 99 (best).
Column Name	Data type	Description
`node_ip`	inet	Node address.
`latency_index`	int	Ranking by mean read latency during the last sampling period.
`conn_id`	text	Unique client connection ID.
`keyspace_name`	text	Keyspace name.
`read_latency`	double	Mean value in microseconds during the last active sampling period for this object/client.
`table_name`	text	Table name.
`total_reads`	bigint	Count during the last active sampling period for this object/client.
`total_writes`	bigint	Count during the last active sampling period for this object/client.
`user_ip`	inet	Client origin address.
username	text	Present if authentication is enabled.
`write_latency`	double	Mean value in microseconds during the last active sampling period for this object/client.

`object_user_write_io_snapshot` table

Per node recent read/write metrics by client, keyspace, and table during the most recent sampling period. Tracks best-worst latency on a scale of 0 to 99 (0 is worst).
Column Name	Data type	Description
`node_ip`	inet	Node address.
`latency_index`	int	Ranking by mean write latency during the last sampling period.
`conn_id`	text	Unique client connection ID.
`keyspace_name`	text	Keyspace name.
`read_latency`	double	Mean value in microseconds during the last active sampling period for this object/client.
`table_name`	text	Table name.
`total_reads`	bigint	Count during the last active sampling period for this object/client.
`total_writes`	bigint	Count during the last active sampling period for this object/client.
`user_ip`	inet	Client origin address.
username	text	Present if authentication is enabled.
`write_latency`	double	Mean value in microseconds during the last active sampling period for this object/client.

Leases table

Acquire, disable, renew, and resolve are the four lease operations. Histogram statistics indicate the rough distribution of timing: the average amount of time, the time for the worst request out of 100, the absolute worst request, and the rate at which operations are happening.

Lease metrics for the lease subsystem.
Column Name	Data type	Description
`acquire_average_latency_ms`	bigint	Average latency, in milliseconds, to acquire lease.
`acquire_latency99ms`	bigint	Latency recorded 1% from the end, for the 99th percentile.
`acquire_rate15`	double	For the datacenter.
`disable_average_latency_ms`	bigint	The average amount of time.
`disable_latency99ms`	bigint	The time for the worst request out of 100.
`disable_max_latency_ms`	bigint	The absolute worst request.
`disable_rate15`	double	The rate at which operations are happening.
monitor	inet	The machine partially responsible with the lease, there are `Replication_Factor` # of monitors.
name	text	The lease name.
`renew_average_latency_ms`	bigint	The average amount of time.
`renew_latency99ms`	bigint	The time for the worst request out of 100.
`renew_max_latency_ms`	bigint	The absolute worst request.
`renew_rate15`	double	The rate at which operations are happening.
`resolve_average_latency_ms`	bigint	The average amount of time.
`resolve_latency99ms`	bigint	The time for the worst request out of 100.
`resolve_max_latency_ms`	bigint	The absolute worst request.
`resolve_rate15`		The rate at which operations are happening.
up	boolean	Whether the lease is held, implies that the service is up.
`up_or_down_since`	timestamp	Time of the last change. For example, `UP` since 10PM or `DOWN` since 4PM.