Performance Considerations
The performance of real-time data in Tracking Server clients—ArcGIS Tracking Analyst and Tracking Viewer—depends on many factors and can be described in two primary ways: data timeliness and data throughput.
There are several factors that have to be considered when determining Tracking Server performance. Implementations will vary according to the specific characteristics of the environment. The following factors affect performance:
- Message size
- Number of updates per second
- Number of services
- Bandwidth
- CPU
- Memory (RAM)
Data Timeliness
Data timeliness describes the time delay from the event occurrence to the point at which it is displayed on the map by a client. Factors such as network speeds, connection types, and machine performance affect data transmission and processing speed. Data timeliness is also affected by data throughput.
Data Throughput
Data throughput describes how many items can be tracked. The number of tracked items depends first on the data and second on machine and network performance. The first question to ask when evaluating throughput is, How many items do I want to track? The second question to ask is, How often do I want to send an event message? If you are tracking 100 trucks that send a message every second, the data throughput is 100 events per second. If you are tracking 100 trucks that report back every 10 seconds, the data throughput is about 10 events per second. In these two examples, the throughput drastically changes based on the change in message frequency.
What data throughput can Tracking Server, ArcGIS Tracking Analyst, and the Tracking Viewer handle? This part of the performance equation is based on machine and network performance, which is greatly affected by processor speed and memory. Both Tracking Server and its tracking clients store and process events in memory, so an increase in processor power and machine memory will benefit those who need higher data throughput rates. Faster network cards and infrastructure will also benefit those who need higher data throughput.
Bandwidth is the primary constraint to Tracking Server performance at this time. Note that the number of concurrently connected clients will affect bandwidth. As you plan for your system architecture, you will need to take the following into consideration:
How big are the messages?
A typical GPS location report generates a very small message, usually less than 400 bytes.
How often do the items you are tracking report?
In many situations, once per 30 seconds, or even once per minute, is sufficient for the tracking need. You can tune these settings in your GPS units, sensors, or other devices to ensure your throughput needs will be met. Of course, your overall business needs drive this as well.
How many objects will you be tracking?
If you are tracking a fleet of 3,000 trucks, it will require more throughput than if you are tracking only 20.
How many clients will be monitoring the data simultaneously?
Remember to include desktop as well as Internet clients.
Taking these items into consideration, you can do the math to find out what your limitations will likely be. In this example, you have GPS units reporting very simple messages only once per minute.
Network Type: | 100BaseT |
Network Bandwidth: | 100Mbit/second * (8 bits/byte) = 12500 bytes/second |
Adjusted Bandwidth: | 12500 bytes * 80% = 10000 bytes/second potential throughput (to adjust theoretical network throughput for average realities) |
GPS Units: | 5 |
GPS Reporting Rate: | 1 update per minute |
GPS Msg Size: | 200 bytes |
Rate per Minute: | 200 bytes * 5 units = 1000 bytes/min = 16.67 bytes/sec (round to 17) |
Max # of Clients: | 10000 Bytes/17 Bytes = 588 Potential Simultaneous Client Connections |
Client-Related Performance Factors
From the client point of view, ArcGIS Tracking Analyst and Tracking Viewer are designed to serve different levels of use. ArcGIS Tracking Analyst is a GIS-based client with advanced display, query, and analysis capabilities, while Tracking Viewer is a lightweight client with basic display capabilities. Because of its advanced functionality and resultant memory usage, ArcGIS Tracking Analyst may not be best suited to applications that require heavy data throughput.