If you want to see how different protocols such as TCP or UDP perform when it comes to maximizing your network capacity, it’s important to test network throughput appropriately and with the correct mode engaged.
TCP (Transmission Control Protocol) is a reliable protocol used most often for emails and web traffic. UDP (User Datagram Protocol) is the protocol most organizations rely upon when speed or real-time network access is the most important priority (such as for video streaming and for file transfer acceleration).
iPerf Testing of File Transfer Speeds with TCP vs UDP
Testing network throughput with iPerf (an opensource command line tool) can determine network speed and functionality for both TCP and UDP. It’s important to note that this tool will produce standardized performance measurements, regardless of the network.
Some organizations might prioritize their network infrastructure for TCP only, other organizations might allow both TCP and UDP with equal priority, another organization might deny UDP traffic all together, and finally other organizations might give more priority to UDP and throttle down TCP to prevent (denial-of-service-attacks) DoS cyber-attacks.
Before any testing with iPerf can begin, it is important to know what your organization’s network policy is towards TCP and UDP, as well as be familiar with what other organizations may decide on this topic.
The fact that iPerf in TCP mode shows better results than UDP doesn’t necessarily mean that UDP will be slower for file transfers. For example, a long-distance file transfer between New York City and Mumbai over TCP with latency around 220 ms will get a maximum file transfer speed around 2Mbps, while the same file transfer over an accelerated file transfer solution, such as Fortra’s FileCatalyst' UDP-based protocol, can go as fast as 100Mbps.
In this case, any speed over 2Mbps will be better over UDP than TCP. If you are building a file transfer system for long-distance file transfers this is an important factor to consider during your iPerf testing.
Related Reading: Overcoming Distance and File Size When Transferring Files
When using iPerf to test UDP (-u or -udp switch), you must pay particular attention to Lost Datagrams. The iPerf tool might show a very good throughput over UDP but when you look more closely at the results, the lost datagrams (percentage) will be a very high number. This means that your real “goodput” is much lower than what the iPerf tool reports. Re-run your UDP test with lower speed until your lost datagrams are around 1% or less.
When testing accelerated file transfer solutions, such as FileCatalyst, with its UDP-based file acceleration protocol, you may see much better speed results if you add real network latency between the sender and the receiver.
Benefits of Using UDP with iPerf
- Knowing the exact “good put”: With UDP sending data in packets, without the need for acknowledgement, available bandwidth is maximized. Testers must consider Lost Datagrams in the report.
- Network latency does not affect throughput: UDP gets to work without needing to establish a connection before sending data, delivering better speed than TCP over longer distances.
- Ability to specify bandwidth: When testing with iPerf, you can specify bandwidth limits with UDP to better represent network conditions.
- Packet loss measurement: To assess how reliable a network is in specific applications, with UDP you can easily measure any loss by looking at the Lost Datagrams column in the report.
What Speeds are Possible with UDP-based File Transfer Solutions?
- As TCP can be highly susceptible to network latency and packet loss, TCP goodput is always lower than the actual available line speed, even on a relatively stable network. For example, FTP transfers can peak at only 30 Mbps on a 50 Mbps link with packet loss of 0.1% and a delay of 10 ms.
- UDP-based FileCatalyst however, yields goodput of 48 Mbps, which is just slightly less than the maximum available line speed. Even when network conditions deteriorate to 2% packet loss and a delay of 150 ms, transfers can be expected to perform at 450 Kbps, or 1% of the actual available bandwidth. FileCatalyst maintains its 48Mbps goodput.
- You can calculate how fast a specific file transfers with this File Speed Calculator. Just plug in your file size and destination. With FileCatalyst’s UDP-based technology, you’ll see speeds not possible with other file transfer solutions.
In conclusion, iPerf is a great tool to test and compare the TCP and UDP performance. Faster TCP results do not necessarily mean that FileCatalyst will be slower than TCP. The UDP tests with iPerf, combined with Lost Datagrams results under 1% are a great way to establish the maximum goodput limit for file transfer acceleration tools such as FileCatalyst.
Test FileCatalyst’s Transfer Speeds Yourself
If you need to get large files to their destination quickly and reliably, FileCatalyst, with its UDP-based technology, delivers. With unmatched speed, FileCatalyst ensures a constant flow of data, thanks to automatic retransmission in case of a lost packet of data. Try it risk-free for yourself today.