This document describes how to manage database parameters for TDSQL Boundless read-only analysis instances, including viewing and modifying parameters, as well as the meaning and value range of each parameter.
2. In the instance list, click the Instance ID of the target read-only analysis instance to go to the Instance Management page.
3. Select the Parameters Settings tab to view all configurable parameters for the current instance.
Modifying a Single Parameter
1. On the Database Parameters tab, locate the parameter you need to modify.
2. Hover the mouse over the Current Value column and click the
that appears.
3. Enter the new value in the pop-up input box and click
to save.
Modifying Parameters in Batches
1. On the Database Parameters tab, click Batch Modify Parameters in the upper-left corner.
2. In the pop-up panel, select and modify the values of multiple parameters.
3. Click Confirm Modification to submit the modifications.
Note:
Before modifying parameters, carefully read the Acceptable Values. Values outside the allowed range will be rejected. If the modified parameters require an instance restart to take effect, it is recommended that you perform the operation during off-peak hours and ensure that your application has a reconnection mechanism.
Parameter List
Data Loading Parameters
Performance parameters for controlling data synchronization from TDSQL Boundless read/write instances to read-only analysis instances.
Parameter
Default Value
Modifiable Range
Restart Required
Parameter Modification Recommendations
replication_sink_parallelism
16
[1 - 64]
No
Modifying this parameter increases the concurrency during incremental data loading. This can speed up incremental data loading but will increase the load on the read/write instance. Adjust this parameter when the read/write instance has sufficient resources.
replication_dumper_table_parallel
2
[1 - 1024]
No
This parameter controls the number of tables that can start full data loading simultaneously. When the read/write instance has sufficient resources, a greater number of tables starting simultaneously results in faster full data loading. Adjust this parameter when the read/write instance has abundant resources.
replication_dumper_insert_parallel
4
[1 - 1024]
No
This parameter controls the write concurrency to the read-only analysis instance during full data loading. When the read-only analysis instance has sufficient resources, a higher write concurrency results in faster full data loading. Adjust this parameter when the read-only analysis instance has abundant resources.
replication_dumper_select_parallel
16
[1 - 64]
No
This parameter controls the number of concurrent threads for reading table data during full data loading. When the read/write instance has sufficient resources, a higher number of concurrent threads results in faster full data loading. Adjust this parameter when the read/write instance has abundant resources.
replication_dumper_chunk_size
10000
[1000 - 5000000]
No
The size of the chunk read from the read/write instance during each full data load. A larger chunk results in faster full data loading but consumes more resources from the read/write instance. Adjust this parameter based on the actual resource usage of the read/write instance.
Compute Engine Parameters
Used to control the parsing, execution, optimization, and runtime behavior of SQL in the columnar read-only analysis engine.
Query Behavior Control
Parameter
Default Value
Modifiable Range
Restart Required
Parameter Modification Recommendations
sql_mode
Consistent with the read-write instance
Any combination of 23 optional values. For details, see the description below.
No
This parameter controls the execution behavior of SQL in the columnar read-only analysis engine, and its behavior is consistent with MySQL. However, some SQL_MODE settings may be incompatible or configurable but ineffective.
max_execution_time
0
[0 - 86400000]
No
This parameter controls the maximum execution time allowed for SQL statements in the unit of milliseconds. SQL statements exceeding this execution time will be automatically interrupted. The default value is 0, indicating that execution will not be interrupted.
libra_max_delay_allow_query
86400
[1 - 31536000]
No
This parameter controls the maximum delay allowed for objects involved in SQL execution within the columnar read-only analysis engine in the unit of seconds. If this time is exceeded, an error indicating excessive delay will be reported when SQL is executed in the columnar read-only analysis engine.
libra_slow_log_threshold
300
[300 - 9223372036854775807]
No
Slow query threshold. SQL statements whose execution time exceeds this value will be recorded in the slow log in the unit of seconds.
libra_query_log_max_len
4096
[4096 - 4194304]
No
This parameter controls the maximum length of SQL records in slow log management in the unit of KB.
group_concat_max_len
1024
[4 - 4294967295]
No
The maximum length of the result returned by the GROUP_CONCAT() function in the unit of Byte.
Controls enabling/disabling the Runtime Filter feature.
libra_runtime_filter_type
MIN_MAX,BLOOM_FILTER,IN_FILTER,TOPN_FILTER
Any combination of 4 optional values
No
Indicates the settings of the Runtime filter types that can be assigned.
libra_enable_late_materialization
ON
[ON | OFF]
No
Controls enabling/disabling the late materialization feature.
libra_enable_snapshot_read
ON
[ON | OFF]
No
Controls whether to enable the snapshot read feature for the columnar read-only analysis engine.
libra_preserve_order_for_pagination
OFF
[ON | OFF]
No
This parameter controls whether to globally enable the pagination order-preserving feature.
libra_max_mpp_query_num
0
[0 - 1000000]
No
This parameter controls the number of SQL statements that can be concurrently executed by the columnar read-only analysis engine. The default value is 0, which means no limit. It is usually enabled only to prevent excessive concurrency from causing SQL statements to consume all database resources. After it is enabled, subsequent SQL statements will be rejected for execution if the allowed number is exceeded.
ETL-Related
Parameter
Default Value
Modifiable Range
Restart Required
Parameter Modification Recommendations
libra_concurrent_etl
OFF
[ON | OFF]
No
Controls whether to enable concurrent writeback to the read-write instance.
libra_etl_concurrency
1
[1 - 128]
No
Controls the number of concurrent threads for writeback to the read-write instance. It takes effect only when libra_concurrent_etl is set to ON.
libra_etl_to_rw
OFF
[ON | OFF]
No
Controls whether to enable the accelerated ETL writeback feature.
Storage Engine Parameters
Used to control the resource configuration of underlying storage, such as block size, number of threads, and cache capacity.
Parameter
Default Value
Modifiable Range
Restart Required
Parameter Modification Recommendations
max_block_size
65409
[0 - 4294967295]
No
This parameter controls the size of a single data block during vectorized execution, in MB. A larger value may lead to excessive scanning costs in scenarios with fine-grained queries. Conversely, a value that is too small results in a large number of block scans, which can degrade performance. Tuning this parameter primarily depends on the query scenario. If most queries involve large-scale data scanning and computation, consider setting this value higher. If most queries are simple point lookups or involve small result sets, consider setting this value lower.
max_threads
16
[0 - 1024]
No
This parameter controls the maximum number of threads that SQL can use in the columnar read-only analysis engine. By default, it is consistent with the maximum CPU of the instance specification. It is recommended to modify it to the instance specification's CPU/2.
runtime_filter_max_in_num
1024
[0 - 4294967295]
No
Maximum NDV value allowed when an IN-type Runtime Filter is generated.
block_cache_capacity_mb
512
[0 - 2147483647]
Yes
This parameter controls the cache size of the read-only analysis instance, in MB. A larger cache space can cache more table data to improve execution efficiency after multiple SQL executions and avoid repeated data scanning from the disk.
This parameter requires an instance restart to take effect after modification.
