V-PERF is VeEX's implementation of IETF RFC6349's stateful TCP/UDP upload and download throughput (speed) performance tests for Ethernet links. V-PERF is compatible with iPerf3.
The RFC6349 test framework describes a practical standardized methodology that uses iPerf to measure stateful TCP throughput, end-to-end, in a managed IP network. The goal is to provide a better indication of the expected user experience for business services. It is also used to optimize the TCP and IP configuration parameters for improved latency and throughput. The iPerf test is also being adopted as more controlled and reliable speed tests to validate Internet access services' Quality of Experience (QoE) for small business and residential troubleshooting applications.
The RFC6349 test methodology, from IETF, helps service providers optimize their network for transmission control protocol (TCP) traffic and maximize its performance across a broad set of applications. It stresses networks with true stateful TCP test traffic, to assess sustaining TCP performance on the current network/link configuration, using multiple streams in parallel. That is much closer to the subscribers' experience and required by service providers for QoE and SLA verification.
iPerf, from the Internet Performance Working Group, is an open-source speed test and network performance measurement tool. It uses multiple TCP and UDP streams to send traffic from one host to another. It helps determine the network's maximum throughput and top bandwidth speed. It can also help optimize TCP windows for optimal performance. iPerf is part of the RFC6349 test methodology.
V-PERF TCP/UDP upload and download Layer 4+ performance is a hardware-based benchmark test addressing common sources of customer complaints that come from actual file transfer speeds not matching the throughput rates guaranteed by service providers in their Service Level Agreements (SLA). While many factors can affect TCP applications performance, including customer’s operating system hardware performance and settings (e.g., TCP window size), carriers need to prove their SLA with a test tool that can show the TCP performance independent of Operating System or far-end Server limitations, and present repeatable reliable results.
The test sets' V-PERF feature uses iPerf, part of the RFC6349 test methodology and metrics, for qualifying network TCP or UDP performance. It offers a full line rate stateful TCP test with configurable window sizes, client and server modes, as well as compatibility with iPerf servers.
- TCP/UDP Throughput Compliant with the RFC6349 test methodology
- Stateful TCP/UDP Test at line rate
- TCP/UDP Client and Server modes
- Compatible with iPerf and iPerf3 Client/Server
- Up to 64 parallel test streams
- MTU search per RFC4821
- Round Trip Time Measurement (RTD)
- Configurable TCP Window sizes. with Manual and Auto window sizing
- Multi-Window size tests
- Measurements: TCP/UDP Throughput rate (min, max, average), Transfer file size and duration, Transfer time ratio, Efficiency %, Buffer Delay %
- Test Duration configurable by Time or File Size
- Note: These features and capabilities may vary by test set and/or test interface.
V-PERF or iPerf TCP/UDP testing are more robust throughput tests for the field (access network or link), performed against centralized iPerf servers or dedicated test heads (e.g., 10GE or 100GE RTU-300+), to evaluate the network's Internet performance for business customers, providing reports with more technical details and results. V-PERF is a set of hardware-based iPerf test clients and centralized RTUs (Remote Test Unit) test heads, capable of handling more than 100 field test sets simultaneously, without any drop in performance (using up to 100GE connections at the node). V-PERF is also used to periodically run scheduled test to verify the interconnect links between nodes.
V-PERF is also compatible with existing iPerf3 applications running on servers. However, users adopting this approach must be aware of their (servers') CPU-based limitations under stress.
Although these professional tools are not intended for residential/consumers, since they require equipment installed in the operators' or service provider's network, they are a must for technicians troubleshooting internet access issues. They should bring these tools when installing, activating, validating or troubleshooting modern multi-gigabit services. Centralized remote test units (RTU) are also used within the networks' interconnect, to verify their own nodes' interconnection links and serve as traffic generators for multiple field test sets. For business services, customers’ IT groups can also set their own iPerf3 servers to test leased lines between their points of presence, to monitor, diagnose and troubleshoot QoS issues.
TCP vs. UDP Test Traffic
To fully understand stateful TCP throughput (the heart of V-PERF/iPerf), it may require many years of studying networking, telecommunications, congestion control protocols, and trial and error with network emulating software, network impairment equipment, and actual tests with test sets and network equipment. However, that is not the goal of this article, so we should at least understand that UDP traffic is not guaranteed, and TCP traffic is guaranteed. Because of its reliable nature, TCP traffic can experience different performance characteristics based on different network loads, conditions, congestions, etc.
TCP Reliability
Although TCP is preferred for data transmission, the network latency, packet loss, jitter, buffering (from network equipment) link/network characteristics all contribute to the good or bad performance of TCP traffic. Network latency may be one of the biggest issues with TCP performance.
The Effects of Latency on TCP Throughput Performance
As a simple example, focusing on bursts of packet loss and increasing delay, we can see the performance effects they have on TCP throughput. The graph below is a representation of a TCP throughput test that is running at its maximum performance of 4.7 Gbit/s approximately, at the start of the test. As the test progresses, we introduce a burst of ten lost packets at a given network latency. The smaller the latency, the faster the TCP throughput recovers to its max performance. Keeping the same burst of lost frames as the test continues, and increasing delay, you can see that the "slow start" process, takes longer. This means that the recovery time to max performance is increased as network latency increases. Now, imagine the complexities of more realistic cases, in which the latency varies greatly due to window buffering, variable packet loss, jitter, etc.
Related Test Solutions
- Ethernet Test Solutions page
- CX310 - DOCSIS 3.1 Installation Tool
- CX380C - Advanced DOCSIS & Cable Plant Maintenance Test Set
- FX120 - XGS-PON Analyzer (for Multi-Gigabit Internet Access Services)
- MTTplus-340 - 10G Multi-Service Test Module
- MTX150 - 10G Multi-serviceInstallation & Maintenance Test Set
- MTX150x - 10G Ethernet Services Installation Test Set
- MTX150x Lite - Multi-Gigabit Internet Services & Ethernet Speed Test Solution
- PX90 - PON Service Installation Meter
- RTU-340 - Advanced 10G Remote Test Unit
- RTU-600x - Advanced 100G Remote Test Unit
- RXT-6802 - Advanced 800G Multi-Service Test Module
- RXT-6402 - Advanced 2x400G Multi-Service Test Module
- RXT-6400 - Advanced 2x400G Multi-Service Test Module
- RXT-6200 - Advanced 100G Multi-Service Test Module
- RXT-3400 - Advanced 10G Multi-Service Test Module
- TX300s-100Gx - Advanced 100G Multi-Service Test Set
- TX340s - Advanced 10G Multi-Service Test Set