Instance Parameters
By default, DDM allows you to modify only the following parameters. If you need to modify other parameters in some special scenarios, such as data migration, contact technical support.
Parameter |
Description |
Value Range |
Default Value |
---|---|---|---|
bind_table |
Data association among multiple sharded tables. The optimizer processes JOIN operations at the MySQL layer based on these associations. For details about parameter examples, see the description below the table. |
- |
|
character_set_server |
DDM server's character set. To store emoticons, set both this parameter and the character set on RDS to utf8mb4. For a DDM instance 3.0.9 or later, you can execute show variables like '%char%' to query its character set. You will find that character_set_client, character_set_results, and character_set_connection in the command output all have a fixed value, utf8mb4. |
gbk, utf8, utf8mb4 |
utf8mb4 |
collation_server |
Collation on the DDM server. |
utf8mb4_unicode_ci, utf8mb4_bin, utf8mb4_general_ci |
utf8mb4_unicode_ci |
concurrent_execution_level |
Concurrency level of scanning table shards in a logical table:
|
RDS_INSTANCE, DATA_NODE, PHY_TABLE |
DATA_NODE |
connection_idle_timeout |
Number of seconds the server waits for activity on a connection before closing it. The default value is 28800, indicating that the server waits for 28,800 seconds before closing a connection. |
60~86400 |
28800 |
contains_shard_key |
Whether the SELECT, UPDATE, and DELETE statements must contain sharding keys in filter conditions. |
OFF or ON |
OFF |
ddl_precheck_mdl_threshold_time |
Threshold of the MDL duration in DDL pre-check. The unit is second. The default value is 120. |
1~3600 |
120 |
enable_table_recycle |
After the table recycle bin is enabled, deleted tables are moved to the recycle bin and can be recovered by running the RESTORE command within seven days. |
OFF or ON |
OFF |
long_query_time |
Minimum duration of a query to be logged as slow, in seconds. The default value is 1, indicating that the query is considered as a slow query if its execution duration is greater than or equal to 1 second. |
0.01~10 |
1 |
max_allowed_packet |
Maximum size allowed for the packets transferred between the server and the client at a time. The value must be a multiple of 1024. |
1024~1073741824 |
1073741824 |
max_backend_connections |
Maximum of concurrent client connections allowed per DDM instance. When this parameter is set to 0 (default), the maximum concurrent connections from a DDM node to an RDS instance is: (RDS instance's maximum connections - 20)/DDM nodes. This parameter does not take effect only after maximum connections are set on RDS. |
0~10000000 |
0 |
max_connections |
Minimum concurrent connections from a DDM instance node to the client. This value depends on node classes and processing capabilities of the target data node. Too many connections may cause connection waiting, affecting performance. The consumption of DDM connections varies with the number of shards and SQL design. For example, If a SQL statement contains a sharding key, each DDM connection consumes one data node connection. If the SQL statement contains no sharding keys and the number of shards is N, N data node connections are consumed. If SQL design is appropriate and processing capabilities of DDM and its data nodes are good enough, you can set this parameter to a value slightly smaller than the product of backend data nodes x maximum connections supported by each data node. Carry out pressure tests on your services and then select a proper value. |
10~40000 |
20000 |
min_backend_connections |
Minimum of concurrent client connections allowed per DDM instance. The default value is 10. |
0~10000000 |
10 |
seconds_behind_master |
Threshold in seconds of the replication lag between a primary RDS instance to its read replica. The default value is 30, indicating that the time for data replication between the primary RDS instance and its read replicas cannot exceed 30 seconds. If the time exceeds 30 seconds, the data read requests are no longer forwarded to the read replicas. |
0~7200 |
30 |
sql_execute_timeout |
Number of seconds to wait for a SQL statement to execute before it times out. The default value is 28800, indicating that the SQL statement times out if its execution time is greater than or equal to 28,800 seconds. For data nodes, ensure that net_write_timeout has a greater value than sql_execute_timeout. |
100~28800 |
28800 |
temp_table_size_limit |
Size of a temporary table. |
500000~2000000000 |
1000000 |
transfer_hash_to_mod_hash |
Whether the hash algorithm must be converted into mod_hash during table creation. |
OFF or ON |
OFF |
ultimate_optimize |
Whether the SQL execution plan is optimized based on parameter values. |
OFF or ON |
ON |
force_read_master_in_transaction |
Whether SQL statements involved in each transaction are read from the master node.
CAUTION:
This parameter is available in version 3.0.9 or later. If this feature is enabled in version 3.0.9 but the version is downgraded to 3.0.9 below, the feature keeps enabled when the version returns to 3.0.9 or later. |
OFF or ON |
OFF |
ddl_flowcontrol_threshold |
Maximum number of DDL executions (including successful and failed executions) allowed per hour. |
1000~100000 |
10000 |
Feedback
Was this page helpful?
Provide feedbackThank you very much for your feedback. We will continue working to improve the documentation.See the reply and handling status in My Cloud VOC.
For any further questions, feel free to contact us through the chatbot.
Chatbot