Reasons for high performance connectivity
What is happening
Alongside classic data traffic between client and server, servers in data centers are now communicating more with one another – a result of virtualization. If in this challenging environment a link to a physical server cannot come up with the goods in terms of performance, this naturally has an effect on all virtual machines and applications running on the server.
Bit errors and packet loss can thus have dramatic effects. Alongside routers and switches, cabling is one of the most important factors influencing throughput and latency.
Good to know
The “slowcoaches” in a network often remain undiscovered. Technicians do test their installations to see that they are adhering to the projected network requirements. Individual optical measurements, such as measuring IL and RL, usually capture and integrate the test signal over a period of 300 milliseconds. The optical pulses of a 10, 40 or 100 Gigabit Ethernet application are, however, only 100 picoseconds long – three billion times shorter.
Conventional test methods cannot resolve any optical phenomena, such as reflection or modal noise, that take place at bit level. R&M has examined this problem in its own research and development projects and in stringent benchmark tests and developed appropriate solutions.
Ultimately, anything you can possibly do to avoid bit errors and packet loss, even in terms of network connectivity, must be done. R&M pursues this goal with its High Performance Network Connectivity (HPNC) range. R&M has proved that it is possible to transmit without any packet loss. The corresponding cabling is even relatively inexpensive to install. For more background information, read the specialist article in CONNECTIONS no. 48.
Help us improve. Send your feedback and suggestions to firstname.lastname@example.org