Neural network resource sizing apparatus for database applications

• US 8,051,019 B2
• Filed: 07/13/2006
• Issued: 11/01/2011
• Est. Priority Date: 07/13/2006
• Status: Active Grant

• Current Assignee: SAP SE
• Original Assignee: SAP AG (SAP SE)
• Inventors: Sigal, Lev, Ziv, Shai
• Primary Examiner(s): Gaffin; Jeffrey A
• Assistant Examiner(s): Wong; Lut
• Application Number:
  US11/486,428
• Publication Number:
  US 20080016014A1
• Time in Patent Office: 1,937 Days
• Field of Search: 706/21, 709/226, 713/100
• US Class Current: 706/21
• CPC Class Codes: G06N 3/02 Neural networks

Abstract

A neural network resource sizing apparatus for database applications. Through use of multiple database application metrics input into a neural network learning algorithm, recommended resource capacities are generated. Input parameters such as the number of records, lookups, images, PDFs, fields, BLOBs and width of fields for example may be utilized to train a neural network to yield needed resource metrics such as the processing power, memory, disk and/or network capacities required to run the database application. Training for the neural network may involve running tests over all desired cross interactions of input and output parameters beginning for example with a small repository and ending with the maximum complexity of data and schema test. The training data is input into the neural network for the given database application version and utilized to plan resource utilization. A portal or webservice may be utilized to provide an interface to the apparatus.

Claims

1. A non-transitory, computer-readable medium storing program code executable by a computer to perform a method, said method comprising:

initiating, by a test server, a neural network configured to recommend resource purchases for a desired installation;

automatically generating, by the test server, a plurality of resource output results, wherein each resource output result is generated by automatically loading a test database application with a test schema and automatically running a performance load on said test database application loaded with said test schema;

automatically modifying, by the test server, said neural network by training said neural network using said plurality of resource output results as training data;

automatically obtaining, by the test server, at least one database application parameter for said desired installation, wherein the test server is able to process multiple types of database application parameters including all of;

a number of database records, a number of database lookups, a number of images in a database, a number of BLOBs, a number of PDF files stored in a database, a number of database fields, and a width of a plurality of database fields; and

automatically obtaining, by the test server, at least one recommended resource output result from said neural network based on said at least one database application parameter, wherein said at least one recommended resource output result is a recommended resource purchase for said desired installation.

2. A non-transitory, computer-readable medium storing program code executable by a computer to perform a method, said method comprising:

preparing, by a test server, a neural network in an initial state, wherein said neural network is configured to recommend resource purchases for a desired installation;

automatically generating, by the test server, a first training performance point comprising resource utilization information, wherein said first training performance point is generated by automatically loading a test database application with a first test schema and automatically running a first performance load on said test database application load with said first test schema;

automatically generating, by the test server, a second training performance point comprising resource utilization information, wherein said second training performance point is generated by automatically loading a test database application with a second test schema and automatically running a second performance load on said test database application load with said second test schema;

automatically modifying, by the test server, said neural network by training said neural network using said first training performance point and said second training performance point;

automatically obtaining, by the test server, at least one database application parameter for said desired installation, wherein the test server is able to process multiple types of database application parameters including all of;

a number of database records, a number of database lookups, a number of images in a database, a number of BLOBs, a number of PDF files stored in a database, a number of database fields, and a width of a plurality of database fields; and

automatically obtaining, by the test server, at least one recommended resource output result from said neural network based on said at least one database application parameter, wherein said at least one recommended resource output result is a recommended resource purchase for said desired installation.

3. The medium of claim 2 wherein said first performance point and said second performance point each comprise a central processing unit benchmark number and unit of measure associated with database computer hardware.

4. The medium of claim 2 wherein said first performance point and said second performance point each comprise an amount of random access memory associated with database computer hardware.

5. The medium of claim 2 wherein said first performance point and said second performance point each comprise an amount of disk storage space for database hardware.

6. The medium of claim 2 wherein said first performance point and said second performance point each comprise a network throughput speed.

7. The medium of claim 2 wherein said neural network is a perceptron.

8. The medium of claim 7 wherein said perceptron utilizes a hard-limit transfer function.

9. The medium of claim 7 wherein said perceptron comprises a multi-layer neural network.

10. The medium of claim 7 wherein said perceptron uses a learning rule formula W^new=W^old+ep^T and b^new=b^old+e, where e=t−

a.

11. The medium of claim 2 further comprising a website portal for providing said at least one recommended resource output result.

12. The medium of claim 2 further comprising a website portal wherein said website portal is configured to provide said at least one recommended resource output result via a webservice XML response.

13. The medium of claim 2 further comprising a website portal configured to obtain ongoing output results from at least one customer installation over a period of time.

14. The medium of claim 2 further comprising a website portal configured to obtain ongoing output results from at least one customer installation and report a high utilization to a customer.

15. A method, comprising:

initiating, by a test server, a neural network configured to recommend resource purchases for a desired installation;

automatically generating, by the test server, a plurality of resource output results, wherein each resource output result is generated by automatically loading a test database application with a test schema and automatically running a performance load on said test database application loaded with said test schema;

automatically modifying, by the test server, said neural network by training said neural network using said plurality of resource output results as training data;

automatically obtaining, by the test server, at least one database application parameter for said desired installation, wherein the test server is able to process multiple types of database application parameters including all of;

a number of database records, a number of database lookups, a number of images in a database, a number of BLOBs, a number of PDF files stored in a database, a number of database fields, and a width of a plurality of database fields; and

automatically obtaining, by the test server, at least one recommended resource output result from said neural network based on said at least one database application parameter, wherein said at least one recommended resource output result is a recommended resource purchase for said desired installation.

16. A system, comprising:

a processor executing a neural network;

a test schema database; and

a test server coupled to the neural network and test schema database, wherein the test server is to;

automatically generate a plurality of resource output results, wherein each resource output result is generated by automatically loading a test database application with a test schema from the test schema database and automatically run a performance load on said test database application loaded with said test schema;

automatically modify the neural network by training the neural network using the plurality of resource output results as training data;

automatically obtain at least one database application parameter for a desired installation, wherein the test server is able to process multiple types of database application parameters including all of;

a number of database records, a number of database lookups, a number of images in a database, a number of BLOBs, a number of PDF files stored in a database, a number of database fields, and a width of a plurality of database fields; and

automatically obtain at least one recommended resource output result from the neural network based on the at least one database application parameter, wherein the at least one recommended resource output result is a recommended resource purchase for the desired installation.