Datasheets __link__ — Filecatalyst

| Network Path | Latency (ms RTT) | Packet Loss | Throughput (Mbps) – 1 stream | Throughput (Mbps) – 10 streams | |--------------|------------------|-------------|------------------------------|--------------------------------| | LAN | 1 | 0% | 940 | 940 (saturated) | | Transatlantic| 85 | 0% | 870 | 940 | | Transatlantic| 85 | 1% | 560 | 810 | | Satellite | 550 | 3% | 210 | 520 |

| Network Condition | FileCatalyst (1 stream) | FTP over TCP (1 stream) | |------------------------|-------------------------|--------------------------| | 100 Mbps, 0% loss, 100 ms | 95 Mbps | 92 Mbps | | 100 Mbps, 2% loss, 100 ms | 62 Mbps | 3 Mbps | | 10 Mbps, 5% loss, 500 ms | 8.5 Mbps | <0.1 Mbps (timeout) | filecatalyst datasheets

| Component | Minimum (100 Mbps) | Recommended (10 Gbps) | |-----------|--------------------|------------------------| | CPU | 2 cores @ 2.0 GHz | 8 cores @ 3.5 GHz (Xeon) | | RAM | 4 GB | 16 GB | | Disk I/O | SATA (80 MB/s) | NVMe RAID (2 GB/s) | | NIC | 1 GbE | 10 GbE with RSS | | Network Path | Latency (ms RTT) |

Abstract FileCatalyst is a proprietary UDP-based file transport protocol developed by IBM (formerly FileCatalyst Inc.) to overcome the throughput limitations of TCP over high-bandwidth, high-latency networks. Its datasheets provide critical engineering specifications for performance modeling, resource planning, and interoperability. This paper dissects the components of a typical FileCatalyst datasheet, explains the underlying protocol mechanics, interprets key performance metrics (throughput, latency tolerance, packet loss recovery), and analyzes hardware/software SKUs across the product line (Server, Workstation, Agent, HotFolder, Central). We conclude with a comparative analysis against TCP-based alternatives and guidance for datasheet-driven procurement. 1. Introduction File transfer acceleration tools are often evaluated via marketing claims. For rigorous engineering evaluation, datasheets remain the authoritative source of specifications. FileCatalyst’s datasheets are unique because they quantify performance under non-ideal network conditions—specifically, packet loss and variable latency. We conclude with a comparative analysis against TCP-based