Browse Prior Art Database

Providing Intelligent Workload Generator To Simulate Realistic Database Cache Utilization

IP.com Disclosure Number: IPCOM000237946D
Publication Date: 2014-Jul-23
Document File: 3 page(s) / 125K

Publishing Venue

The IP.com Prior Art Database

Abstract

A method and system is disclosed for providing a highly Intelligent Workload Generator (IWG) to handle database data and create oscillating windows for specific transaction within logical bands of the database.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 49% of the total text.

Page 01 of 3

Providing Intelligent Workload Generator To Simulate Realistic Database Cache Utilization

The disclosed method and system provides a highly Intelligent Workload Generator (IWG) to handle database data and create oscillating windows for specific transactions

within logical bands of a database. This is helpful because, among other reasons, hardware constraints are often encountered during proof of concept (POC) that require a demanding performance benchmark on a very large database, such as requiring that every single record be accessed in random fashion. The typical work around includes obtaining more hardware by adding extra control units or using Solid State Drives (SSDs). This may be cost prohibitive. Further, the scope of test can be reduced, due to unrealistic high cache hit ratios. Embodiments of the present invention provide granularity of control and ease of use of changing or mixing workloads to maximize utilization of an existing POC hardware setup while maintaining a realistic cache profile.

The logical bands that are provided as described herein for performance tests, can span the entire database, sub-sections of the database, and can also overlap. The bands act as a wall for different transaction types and allow customization. Accordingly, the bands may prevent contention between certain transaction types. The bands can also utilize a built-in, intelligent randomization algorithm for ensuring that every record in each band is hit. The number of transaction types that can run simultaneously is limitless.

The IWG has tooling that can parse and sanitize the data used to populate the database in order to construct the different transaction bands. The transaction bands are sent to a target system. A subset of datasets is imported to a client side of the IWG through a specification via a configuration file. The IWG then constructs a large in-memory table containing all the transaction types to ensure that the bands of transaction types are in correct ratios. The correct ratios can also be specified to the IWG. In addition to the transaction types and transaction ratios, other configurations can be specified to the IWG including window sizes and windows oscillation speed

within the bands. Windows can be a subset of logically continuous transactions that can be randomly accessed within a defined transaction band type.

The IWG utilizes a Poisson distribution algorithm. The Poisson distribution algorithm is an efficient randomization algorithm that performs optimally. The randomization algorithm behaves in a predictable fashion such that tests repeatability is possible based upon an initial seed value. The randomization algorithm ensures that all entries in a sample set are selected, such that no single or a small set of entries are selected frequently. A distribution of selected entries is accomplished in a relatively even distribution. The randomization algorithm ensures that the randomization does not yield an unrealistic ca...