Browse Prior Art Database

Accurate Softswitch High Availability Benchmark Based on Fault Injection

IP.com Disclosure Number: IPCOM000010333D
Publication Date: 2002-Nov-20
Document File: 3 page(s) / 107K

Publishing Venue

The IP.com Prior Art Database

Abstract

The disclosed method is a software package used to benchmark a Softswitch system. The disclosed method produces a single binary metric that describes the percentage of all call attempts made by end users. Benefits include improved accuracy and a method that can easily be accepted as a standard.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 50% of the total text.

Accurate Softswitch High Availability Benchmark Based on Fault Injection

The disclosed method is a software package used to benchmark a Softswitch system. The disclosed method produces a single binary metric that describes the percentage of all call attempts made by end users. Benefits include improved accuracy and a method that can easily be accepted as a standard.

Background

Currently, a standard Softswitch benchmark system does not exist. Existing methods developed by various TEMs are either lacking accuracy and completeness, or are too complex and difficult to standardize. Therefore, the results of one method cannot be compared with the results of another.

General Description

The disclosed method includes a Softswitch call generator, a fault injection test harness, a set of hardware faults testing scripts, and a controlling program. However, this disclosure focuses on the methodology of the system. The methodology is divided into three steps (see Figure 1):

 

  • Measuring availability by stressing. This step activates all software faults as much as possible in order to achieve stable and accurate fault results. The process can be accelerated so that the workload covering two years can be stressed into about 8 months (based on the relation of busy hour call attempts [BHCA] and mean hour call attempts [MHCA]).Also, application and operating system failures are considered (the total number of interruptions caused by software). When a failure occurs, the MTTR is estimated for those failures.

Note. The workload should be designed according to feedback from customers. For example, the distributions of traditional SS7 call attempts, H.323 call attempts, and SIP call attempts can be used. Distributions from customers help simulate the real environment.

Stressed Availability = (Total of successfully accepted call attempts –

Failed call attempts during downtime)  / (Total call attempts)             

                                                                                                   

In addition, we can calculate the failure rate in normal state lnormal that will be used in  the step 2:

                    Normal failure rate lnormal = 1 – (Total of successfully accepted call attempts )

/ ( Total call attempts )

 

  • Measuring availability by fault injection. This step performs fault injection experiments on the standard hardware platform running Standard Linux. Requests are generated at the rate of MHCA. At the same time, various kinds of hardware faults are injected.
    The...