Solving a network problem with SYSTIMAX XL8 GigaSPEED Solution

Anixter stepped up to provide superior cabling solutions for a world-leading power systems provider.

Specialising in designing, manufacturing and supporting a comprehensive range of products and services for air, sea and land applications, this manufacturer has a long heritage and has been servicing both the commercial and military gas turbine engine markets for more than 45 years.

During that time, it has produced over 42 000 engines and supported more than 30 000 gas turbine engines in over 67 nations around the world. Its markets encompass single-engine operators and large fleets, commercial and military aircraft and industrial/marine engines, rotary-wing and fixed-wing applications in rural and urban locations.

The manufacturer began installing its current cabling infrastructure in 1988. At the time, the products chosen and installed by the manufacturer were Category 5 compliant. Between 1988 and 2002, the manufacturer continued to install these same products (these components had been enhanced as the standards had changed). During this same time period, the manufacturer implemented an ATM backbone (the second largest in the world) and a mix of 10 and 100Mbps Ethernet to the desktop.

In 2000, the manufacturer began to experience network problems. New bandwidth-intensive software applications were not operating as efficiently as they should. For example, whenever a CAD file was pulled from the server across the ATM network, common practice dictated that the file should be modified and then put BACK on the file server. Because of the constant delays, engineers would sometimes keep the files at their local PC instead of putting them back on the file server. Any potential problems with the local PCs would then result in the files being lost forever! In an effort to locate the source problem, the manufacturer purchased a state-of-the art network test device. The testing showed serious impedance and return loss issues attributable to the cabling system.

The manufacturer decided to look at its entire network - active and passive components - and determine how to improve and create a more responsive network. With ATM no longer a viable choice for the backbone, the manufacturer decided to implement a Gigabit Ethernet backbone. Hoping to alleviate some of the slow response times and traffic bottlenecks, it decided to implement Fast Ethernet to the workstation. Finally, to make its infrastructure choice, the manufacturer turned to the Anixter Lab to see how its unique network equipment and traffic would run over different cabling options.

The engine manufacturer wanted to compare its current cabling system to one it was considering for a building-wide rewiring project. Two tests were set up. The first looked at the electrical or MHz bandwidth of the cabling system-at parameters such as crosstalk, return loss and attenuation. The second test examined each cabling system`s ability to transport data measured through frame error rates. These two tests combined gave a more complete picture relative to the quality of the solution and its ability to reliably carry data traffic.

To accomplish this goal, two cabling channels were installed and tested. The first represented the manufacturer`s current cabling channel-90m of early generation Category 5e, terminated on a 110 block using 110C4 connecting blocks with 2-pair 110 to RJ45 patch cables connecting into the network equipment. At the workstation, each Category 5e 4-pair cable was terminated onto two Category 5e information outlets. The second cabling channel was a SYSTIMAX GigaSPEED XL8 Solution. This channel consisted of 90m of 2081 Category 6 cable, VisiPatch connecting blocks, 110 to RJ45 GS8E cords, MGS400 information outlets and RJ45-RJ45 GS8E cords.

The cabling systems were installed in accordance with the permanent link model (without the optional consolidation point connector) of the ANSI/TIA/EIA 568-B.1 document. Both channels were connected to a Fluke DSP4000 field cabling tester. The analyser measured 17 different electrical characteristics - return loss and NEXT are shown.

Both channels were tested for their ability to transport data traffic. Each cabling system was placed between two endpoint computers with a third used for monitoring. Files were transferred between the two endpoints over each cabling system. The files were then transferred at 100Mbps per second. The Data content was generated by Ganymede Chariot software, SAP (creating a sales order) and Windows (2MB and 450MB) file transfer were the application scripts tested on the two cabling systems.

The manufacturer`s current cabling system

Installed Category 5e

* 90 m of early generation Category 5/5e compliant cable
* 110 blocks
* 110C4 connecting blocks
* 110-RJ45 patch cord in the telecommunications room
* Information Outlet with 2-pair patch cord at the work area

The manufacturer`s proposed cabling system

SYSTIMAX GigaSPEED XL8 Solution

* 90 m of SYSTIMAX 2081 Series cable
* VisiPatchTM connecting block
* 110-RJ45 patch cord in the telecommunications room
* MGS400 Modular Jack
* GS8E patch cords at the work area

The Anixter Lab regularly aids customers faced with cabling challenges. Whether addressing a network problem or testing a system under consideration, the Anixter Lab has the tools to complete the picture. The Anixter Lab has a unique approach to network testing. It first examines each component individually and then assembles the channel under consideration and measures the bandwidth resources of that cabling channel. The final measure involves replicating customer applications. This three-step methodology gives the most comprehensive view of the cabling system ability - enabling a quality result.

The GigaSPEED XL8 Solution performed significantly better than the Category 5e products that the manufacturer has installed in its Indianapolis manufacturing facility. In comparing these results to the Category 5e standard, the GigaSPEED XL8 Solution provided significantly more headroom over the standard than the installed Category 5e solution.

The wide margins over the Category 5e and Category 6 standards indicate the substantial electrical bandwidth provided by the GigaSPEED XL8 Solution. The margins achieved by the GigaSPEED XL8 Solution are a result of a high degree of electrical tuning among the channel components. The components comprising the existing cabling system do not exhibit the same degree of tuning and in turn, provide less electrical bandwidth. The relatively large margin over the Category 5e standard for the GigaSPEED XL8 Solution and the relatively low margin for the legacy installed system have a direct impact on the ability of the cabling system to efficiently transport data traffic.

The improved return loss and crosstalk performances directly impact the cabling system`s ability to effectively transport data traffic. The installed Category 5e system experienced significant problems while the GigaSPEED XL8 Solution did not. A number of active data tests were conducted - all underpinning this inference.

Ganymede Chariot software (version 4.0) was used to generate the network traffic over each of the two cabling scenarios between two workstations. The type of traffic sent between the two test workstations was configured using user-defined application scripts provided (as part of the Chariot software). The data was transmitted using 100Mbps Fast Ethernet. The test measured throughput (how much sent data is received without error) and response time (how long it takes to transfer the files and retransmit any errors that may have occurred). A separate test console, consisting of a laptop operating on the network, was used to control the test.

The time spent by the GigaSPEED XL8 Solution to transmit the data averaged 82Mbps compared to less than 1Mbps on the Category 5e solution - this represents an improvement in throughput of over 9 000%. Response time was also significantly better on the GigaSPEED XL8 Solution when compared to the Category 5e solution. The second active data test sent two separate 450-Megabyte files over each cabling channel. The data was transmitted using 100Mbps Fast Ethernet. Again, the test measured both throughput and response time.

The GigaSPEED XL8 Solution transferred the two 450MB files at an average rate of 86.8Mbps and accomplished the transfer in approximately 41 seconds. The Category 5e channel transferred the same two files at an average rate of 0.45Mbps and accomplished the transfer in approximately two hours and 12 minutes.

The application scripts used to generate the network traffic were as follows: TCP Windows File Send (Short Connections) - a filesend transaction that uses short connections (ie, it makes a separate connection for each transaction in the script). The manufacturer requested that the test be configured to transmit (20) 2MB files and (2) 450MB files. These file transfers represent the minimum and maximum file sizes transferred within their network.

In conclusion, the SYSTIMAX GigaSPEED XL8 Solution significantly outperformed the legacy cabling system when tested with the Ganymede Chariot simulation software.

The primary difference in the test results is directly attributed to the superior bandwidth characteristics of the GigaSPEED XL8 Solution, versus the manufacturer`s current cabling plant. This fact is especially evident when conducting TCP file transfers between two workstations. TCP is a Layer 4 transport protocol that is connection-oriented. In other words, the connection between the transmitting and receiving stations must be maintained until the message (or messages) to be exchanged by the application programs at each end have been exchanged.

If connections cannot be established due to packets not being received properly (or in a timely fashion), the receiving station requests retransmission of the packets by the transmitting station. Thus, the TCP handshaking that must occur makes TCP packet transport highly sensitive to CRC errors.