Reliability HotWire

Issue 91, September 2008

Hot Topics

Using BlockSim for System Capability Analysis

In highly available systems, additional metrics of system capabilities are often needed for system evaluation. In this article, we will explore an example where BlockSim's throughput analysis capability is used in order to perform such analysis.

Example of System Capability via Throughput Analysis
Assume a signal is distributed through a network to multiple antenna elements. The signal originates at a single node. It is then distributed to 10 transmitters, each connected to 100 elements. This continues as the signal is split out through various components until the components directly connected to the 1,000 antenna elements are reached. As node components fail, only a portion of the total signal power delivered to the antenna elements is lost. However, if the first node fails, all 1,000 elements are lost and there is no signal. In order to quantify the capabilities of the system in terms of its ability to deliver full signal strength, reliability and availability metrics are not enough. Such a highly redundant system is rarely unavailable. However, it often operates at a lower capability. In order to be able to quantify the signal strength, throughput analysis will be utilized.

Assume that a simplified reliability block diagram (RBD) of the system is constructed as shown below.

To represent this system in BlockSim, a multiple block is utilized to represent 10 antenna elements in parallel. Each antenna element has a throughput of 1 per unit time, while the nodes immediately before and after the antenna elements have a throughput of 100, as they "feed" 10 units to each of these 10 multiple blocks. The starting and ending blocks in the main network have a throughput of 1,000, as they feed 100 units to each of the 10 subsystems.

The properties of the antenna elements are shown next (units are in hours).

The properties of the nodes that feed the antenna elements are as follows:

When all components in the system are available, the throughput is 1,000 units per hour. This is the equivalent to full signal strength. If a node is down, it will make the signal output of a set of 100 antenna elements unavailable.

Predictions for 1 year (8,760 hours) are of interest, with results broken down daily. The simulation settings used are shown next.


A plot of the results for the interval throughput is shown, where an interval is a 24-hour period.

Note that in a 24-hour period, the system can process 24,000 units (24*1,000). Converting the throughput to signal strength entails dividing the interval throughput by 24,000. While the availability remains at 100%, the signal strength goes down to 90% after one year.

As can be seen, although the availability of the system is high, the signal strength is not at full capacity. Throughput analysis provides a better understanding of the capability of the system in cases like this.

Copyright 2008 ReliaSoft Corporation, ALL RIGHTS RESERVED