Browse Prior Art Database

Distributed Storage Test System

IP.com Disclosure Number: IPCOM000097620D
Original Publication Date: 2005-Mar-07
Included in the Prior Art Database: 2005-Mar-07
Document File: 2 page(s) / 45K

Publishing Venue

IBM

Abstract

Testing storage products is a very expensive task, requiring many systems to act as host computers in order to create realistic workloads such that the storage system is tested while operating at or near it's maximum bandwidth capability. Each subsequent family of storage products include higher throughput, more I/O ports, more interface options, and new functions. In order to lower the cost of testing storage, it would be advantageous to be able to use less expensive systems in a loosely-coupled configuration, allowing one to test at a high level of storage bandwidth utilization at a lower cost, and to be able to do so from a centralized control point. The subject invention is a storage test system that is designed to operate in a distributed environment with connections to elements on other systems, providing a flexible test environment with a focus on providing maximum throughput to the devices under test. Although it uses a distributed architecture, it has a centralized point of control. The components are computers with software that create storage commands for the storage devices under test to execute, and they are controlled from a common control point using communication software running on workstations.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 53% of the total text.

Page 1 of 2

Distributed Storage Test System

Figure 1

102

106

114

108

108

108

108

108

108

108

114

114

106

104

110

114

106

106

106

114

102

106

114

                                                  Figure 1 is a schematic representation of the invention. It comprises one or more computer systems 108, which are attached via ethernet 104 such that they may communicate using TCP/IP as known in the art. Also attached to the ethernet 104 are one or more workstations 102, which can also communicate with the computer systems 108 via TCP/IP, and are used to control and monitor the results of the test applications running on the remote computers 108. Said workstations 102 are henceforth referred to as COR or Center Of Room monitors. Each computer system 108 contains at least one fibre channel HBA 106. Each HBA 106 is attached via fibre channel as known in the art to at least one storage device under test 114. Said attachment can either be through a switched network (or fabric), directly attached, or in an arbitrated loop configuration. The invention is not limited to fibre channel attached storage devices, but may also attach to SCSI, SAS, iSCSI, SATA, and any other protocol commonly used in the art to attach storage devices to a host computer.

Workstations 102 are used to launch, control, and monitor the exercisers that run on computer systems 108. An initialization file is read which contains information on the remote computers 108. Connections are established via ethernet TC/PIP to the remote computers 108. The remote test application and it's associated control files are transferred to the remote computer. Then the remote test application is launched. A local connection thread is launched, which is described further later. If there are more remote computers to initialize, then the process continues with the next remote computer. If there are no more remote computers to initialize, the process terminates.

The process that initializes the computers 108 is invoked remotely by the COR monitor application that runs on the workstations 102. A configuration file is read that gives details about the units under test 114 and how to access them. The test environment is initialized, and any input parms that are passed to the application at invocation are processed. The application enters the parms that are read and processed into a data structure to be passed to the testcase threads upon creation.The testcase threads are created and the process terminates.

The testcase process proceeds, and device headers that are used in the dataset to be written to the devices under test are initialized. A flag is set in the device structure that controls the testcase so that the initial pass is a write pass. The process then checks for the type of test bei...