Database tuning using a federated machine learning system of a centerless network

1. A computer-implemented method for database configuration tuning, the computer-implemented method comprising:

selecting, by a computer, a set of database nodes having similar data factors in a centerless network of database nodes;

training, by the computer, configuration models corresponding to the set of database nodes using data parallelism;

combining, by the computer, trained configuration models corresponding to the set of database nodes to form a federated configuration model;

determining, by the computer, whether performance indicators corresponding to the set of database nodes are greater than a performance threshold level;

responsive to the computer determining that the performance indicators corresponding to the set of database nodes are greater than the performance threshold level, recommending, by the computer, a database configuration corresponding to the federated configuration model to a new database node; and

joining, by the computer, the new database node to the centerless network.

2. The computer-implemented method of claim 1 further comprising:

identifying, by the computer, a set of data factors corresponding to each database node in the centerless network of database nodes; and

generating, by the computer, a vector for each database node based on the set of data factors corresponding to each database node.

3. The computer-implemented method of claim 2 further comprising:

calculating, by the computer, vector relationships between the database nodes in the centerless network based on the generated vector corresponding to each respective database node; and

generating, by the computer, a node relationship table for each database node in the centerless network listing neighbor database nodes, data factors, and vector relationships corresponding to each respective database node.

4. The computer-implemented method of claim 3 further comprising:

building, by the computer, the centerless network that includes the database nodes organized into a plurality of virtual database node groups, each virtual database node group based on the generated node relationship table corresponding to each respective database node.

5. The computer-implemented method of claim 1 further comprising:

collecting, by the computer, values of data factors corresponding to a database node in the centerless network of database nodes; and

building, by the computer, a configuration model for the database node based on the values of the data factors corresponding to the database node.

6. The computer-implemented method of claim 5 , wherein the data factors include table cardinality, page number, access frequency, index level, leaf page, input/output speed, and system cache.

7. The computer-implemented method of claim 5 further comprising:

running, by the computer, a database trace on the database node in the centerless network of database nodes; and

identifying, by the computer, a performance indicator corresponding to the configuration model of the database node within data of the database trace.

8. The computer-implemented method of claim 7 further comprising:

mapping, by the computer, the values of the data factors corresponding to the database node to the performance indicator corresponding to the configuration model of the database node; and

recording, by the computer, the mapping of the values of the data factors corresponding to the database node to the performance indicator corresponding to the configuration model of the database node for future reference.

9. The computer-implemented method of claim 1 further comprising:

selecting, by the computer, a set of closest neighbor database nodes to the new database node based on vector relationship values listed in node relationship tables corresponding to the database nodes in the centerless network; and

building, by the computer, the federated configuration model based on configuration models corresponding to the set of closest neighbor database nodes to the new database node.

10. The computer-implemented method of claim 1 further comprising:

generating, by the computer, a database configuration recommendation for the new database node based on the federated configuration model; and

applying, by the computer, the database configuration recommendation to the new database node.

11. The computer-implemented method of claim 1 , wherein the computer uses averaging of configuration model parameters to combine the trained configuration models corresponding to the set of database nodes to generate the federated configuration model.

12. The computer-implemented method of claim 11 , wherein the configuration model parameters include configuration model coefficients, configuration model values, configuration model performance feedback data, and configuration model performance labels.

13. The computer-implemented method of claim 1 , wherein the centerless network of database nodes is located in a cloud environment.

14. A computer system for database configuration tuning, the computer system comprising:

a bus system;

a storage device connected to the bus system, wherein the storage device stores program instructions; and

a processor connected to the bus system, wherein the processor executes the program instructions to:

select a set of database nodes having similar data factors in a centerless network of database nodes;

train configuration models corresponding to the set of database nodes using data parallelism;

combine trained configuration models corresponding to the set of database nodes to form a federated configuration model;

determine whether performance indicators corresponding to the set of database nodes are greater than a performance threshold level;

recommend a database configuration corresponding to the federated configuration model to a new database node in response to determining that the performance indicators corresponding to the set of database nodes are greater than the performance threshold level; and

join the new database node to the centerless network.

15. The computer system of claim 14 , wherein the processor further executes the program instructions to:

identify a set of data factors corresponding to each database node in the centerless network of database nodes; and

generate a vector for each database node based on the set of data factors corresponding to each database node.

16. The computer system of claim 15 , wherein the processor further executes the program instructions to:

calculate vector relationships between the database nodes in the centerless network based on the generated vector corresponding to each respective database node; and

generate a node relationship table for each database node in the centerless network listing neighbor database nodes, data factors, and vector relationships corresponding to each respective database node.

17. The computer system of claim 16 , wherein the processor further executes the program instructions to:

build the centerless network that includes the database nodes organized into a plurality of virtual database node groups, each virtual database node group based on the generated node relationship table corresponding to each respective database node.

18. A computer program product for database configuration tuning, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to perform a method comprising:

selecting, by the computer, a set of database nodes having similar data factors in a centerless network of database nodes;

training, by the computer, configuration models corresponding to the set of database nodes using data parallelism;

combining, by the computer, trained configuration models corresponding to the set of database nodes to form a federated configuration model;

determining, by the computer, whether performance indicators corresponding to the set of database nodes are greater than a performance threshold level;

responsive to the computer determining that the performance indicators corresponding to the set of database nodes are greater than the performance threshold level, recommending, by the computer, a database configuration corresponding to the federated configuration model to a new database node; and

joining, by the computer, the new database node to the centerless network.

19. The computer program product of claim 18 further comprising:

identifying, by the computer, a set of data factors corresponding to each database node in the centerless network of database nodes; and

generating, by the computer, a vector for each database node based on the set of data factors corresponding to each database node.

20. The computer program product of claim 19 further comprising:

calculating, by the computer, vector relationships between the database nodes in the centerless network based on the generated vector corresponding to each respective database node; and

generating, by the computer, a node relationship table for each database node in the centerless network listing neighbor database nodes, data factors, and vector relationships corresponding to each respective database node.

21. The computer program product of claim 20 further comprising:

building, by the computer, the centerless network that includes the database nodes organized into a plurality of virtual database node groups, each virtual database node group based on the generated node relationship table corresponding to each respective database node.

22. The computer program product of claim 18 further comprising:

collecting, by the computer, values of data factors corresponding to a database node in the centerless network of database nodes; and

building, by the computer, a configuration model for the database node based on the values of the data factors corresponding to the database node.

23. The computer program product of claim 22 , wherein the data factors include table cardinality, page number, access frequency, index level, leaf page, input/output speed, and system cache.

24. The computer program product of claim 22 further comprising:

running, by the computer, a database trace on the database node in the centerless network of database nodes; and

identifying, by the computer, a performance indicator corresponding to the configuration model of the database node within data of the database trace.

25. The computer program product of claim 24 further comprising:

mapping, by the computer, the values of the data factors corresponding to the database node to the performance indicator corresponding to the configuration model of the database node; and

recording, by the computer, the mapping of the values of the data factors corresponding to the database node to the performance indicator corresponding to the configuration model of the database node for future reference.