All network copy performance tests were performed using FlexTk Version 1.7.10 between two dual-CPU Xeon machines equipped with 2GB of RAM each and interconnected via a Gigabit Ethernet network with Intel single-port network adapters. For network copy tests we have used two completely identical Western Digital SATA 250GB hard disks equipped with 16MB of cache installed in each test server.

Each benchmark was performed twice, first time we have copied data from the local disk to the network share and the second time we have copied the same data from the network share to the local disk. Before each benchmark we have reformatted the destination disk and rebooted both testing servers.

In order to get a better understanding of how different data sets impact the copy performance, we have used numerous 4GB data sets including a set of 100,000 files 40KB each, a set of 10,000 files 400KB each, a set of 1,000 files 4MB each and a set of 100 files 40MB each. In addition, we have used a 6.1GB real-life data set that included about 150,000 files with 1/3 small files, 1/3 medium-sized files and 1/3 large files.

It is clear from the performance results that the FlexTk's copy engine is better in all but one of the performance tests. This performance advantage may be attributed to two reasons. The first one is the multi-streaming nature on FlexTk's copy engine, which is capable of performing multiple data transfers at the same time thus better utilizing the dual-core CPUs. The second reason lies in some basic limitations of modern Gigabit Ethernet cards, which are capable of reaching the maximum possible transfer speed on numerous simultaneous transfers only.

* This performance review has been prepared for information purposes only and we are strongly advice you to make your own performance evaluations using your specific hardware components and datasets.