Bandwidth Requirements
Understanding the network impact of discovery helps you plan scan schedules and avoid disrupting production traffic. This page explains the bandwidth consumed by different discovery methods and how to control it.
PCAnalyser Discovery Bandwidth
A PCAnalyser discovery generates approximately:
| Direction | Data Volume |
|---|---|
| Upload to target | ~120 KB per computer (the PCAnalyser executable and supporting files) |
| Download from target | ~40 KB per computer (the collected inventory data) |
For a subnet of 254 machines, this amounts to approximately 40 MB total -- spread across the duration of the scan.
Controlling Network Impact
Thread Count
The primary control for network impact is the thread count in the XDSL script. This determines how many machines are scanned simultaneously:
| Thread Count | Effect |
|---|---|
| Low (1--5) | Minimal network impact, but discovery takes longer |
| Medium (10--20) | Good balance for most networks |
| High (50+) | Fastest discovery, but may saturate network links |
To adjust the thread count, edit the XDSL discovery template (see Editing Discovery Scripts).
Scheduling
Schedule discovery during off-peak hours (evenings, weekends) to avoid impacting users:
- Stagger different locations at different times
- Avoid running discovery at the same time as backups or other network-intensive operations
Subnet Splitting
For large subnets or slow WAN links, split the IP range into smaller segments and discover each one separately. This reduces the burst of traffic on any single network link.
Bandwidth by Discovery Method
| Method | Relative Bandwidth | Notes |
|---|---|---|
| PCAnalyser | Low--Medium | ~160 KB per machine (upload + download) |
| WMI | Medium | Depends on how many WMI fields are queried |
| Remote Registry | Low | Small amount of data per field |
| SNMP | Very Low | Small UDP packets per query |
| SSH | Low--Medium | Depends on the number of commands executed |
Tips
- Test with a small group first and measure the actual impact before scaling up
- If users report slowness during discovery, reduce the thread count or reschedule to off-peak hours
- For WAN links, place a Collection Server on the remote side to keep discovery traffic local
Related Articles
- Choosing a Discovery Strategy — selecting the right approach for your network
- Scheduling Discovery Scripts — automating scans at optimal times
- Editing Discovery Scripts — adjusting thread count and technologies