# US 2014/0195505 A1 · Sampling transactions from multi-level log file records

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2014/0195505 A1
- Document ID: US20140195505A1
- Status: Published application
- Kind: A1
- Application number: 14/203,727
- Filed: 2014-03-11
- Published: 2014-07-10
- Earliest claimed: 2013-01-03
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Manoj K. Agarwal; Curt L. Cotner; Amitava Kundu
- Web page: https://patentgrep.com/patent/US20140195505A1

## CPC

- G06F 16/254: {Extract, transform and load [ETL] procedures, e.g. ETL data flows in data warehouses}
- G06F 11/3466: {Performance evaluation by tracing or monitoring}
- G06F 11/3476: {Data logging (G06F11/14, G06F11/2205 take precedence)}
- G06F 16/1865: {Transactional file systems}
- G06F 2201/80: Database-specific techniques
- G06F 2201/87: Monitoring of transactions
- G06F 2201/88: Monitoring involving counting
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A log file contains operation records, each operation record is of a certain type, and each operation record is associated with a transaction. A plurality of operation records is read from the log file into a record store. Records of the plurality of operation records of each operation record type are sampled at a predefined sampling rate. Operation records in the plurality of operations records are identified that are associated with completed transactions of which the sampled operation records are associated. The identified operation records are then extracted from the record store into a data store.

## Claims

### Claim 1

A method for sampling transactions from multi-level log file records, a log file contains operation records, each operation record is of a certain type, each operation record is associated with a transaction, the method comprising:
reading a plurality of operation records from the log file into a record store;
sampling records of the plurality of operation records of each operation record type at a predefined sampling rate;
identifying operation records in the plurality of operations records associated with completed transactions of which the sampled operation records are associated; and
extracting the identified operation records from the record store into a data store.

### Claim 2 (depends on 1)

A method in accordance with claim 1, one operation record associated with a transaction is an end-of-transaction record indicating a completed transaction, the method further comprising:
reading one of the plurality of operation records from the log file into a record store;
recording in a transaction-to-record table an association of the operation record and the transaction to which the operation record belongs;
in response to determining that the operation record is a sampled record and the operation record is an end-of-transaction record, extracting all operation records from the record store into the data store that are associated with the transaction based on the corresponding association in the transaction-to-record table;
in response to determining that the operation record is a sampled record and the operation record is not an end-of-transaction record, recording in an extraction-pending table the transaction to which the operation record belongs; and
in response to determining that the operation record is not a sampled record and the operation record is an end-of-transaction record and the transaction to which the operation record belongs is recorded in the extraction-pending table, extracting all operation records from the record store into the data store that are associated with the transaction based on the corresponding association in the transaction-to-record table.

### Claim 3 (depends on 1)

A method in accordance with claim 1, wherein the plurality of operation records read from the log file further comprises a predefined number of operation records before and after a sample set of operations records and the sampled records are selected from the sample set, such that substantially all operation records associated with the transactions with which the sampled operation records are associated are in the plurality of operation records.

### Claim 4 (depends on 1)

A method in accordance with claim 1, one operation record associated with a transaction is an end-of-transaction record indicating a completed transaction, further comprising:
in response to determining that one of the plurality of operation records associated with a transaction with which a sampled operation record is associated is an end-of-transaction record, extracting the identified read operation records associated with the transaction to a data store.

### Claim 5 (depends on 1)

A method in accordance with claim 1, wherein sampling records of the plurality of operation records further comprises randomly sampling a predetermined number of operation records of each operation record type in the plurality of operation records.

### Claim 6 (depends on 1)

A method in accordance with claim 1, wherein sampling records of the plurality of operation records further comprises sampling operation records of each operation record type at a predefined sampling rate associated with the operation record type.