Choosing a Physical Layer Matrix Switch
The first step to selecting the right physical layer matrix switch for your network is to make sure you match the media type of the switch with that of your network links and devices. Datacom Systems matrix switches are available in copper, single mode fiber, and multi-mode fiber.
In-Line or SPAN Matrix Switch?
Choosing an In-Line or SPAN matrix switch depends on how you are currently doing your network analysis and monitoring and ultimately what data you need to see.
Connecting a network device to a network switch's SPAN port is an excellent way to get an overall view of your network. However, pinpointing the source of actual network problems may require an In-Line solution.
In-Line analysis of your network is possible by creating a direct access port in between two network devices, using network taps or In-Line matrix switches. In-Line access allows you to receive all full-duplex network traffic, including all the errors. In contrast, a monitoring device connected to a SPAN port does not see all traffic. Network switches eliminate Layer 1 and some Layer 2 errors, making it impossible to troubleshoot physical layer problems, bad frames, and corrupt packets. See also: Network Taps versus SPAN for additional insight into whether you should tap your links or use an out of band solution.
If you already have network taps deployed in your network, and you'd like to perform roving analysis on mutiple tapped links, an In-Line matrix switch provides an excellent solution to allow you to leverage your existing infrastructure and save money.
There are also several other considerations you must make when choosing whether to use a SPAN matrix switch or an In-Line matrix switch. The following table provides more information:
Matrix Switch Comparison (In-Line vs. SPAN)
|
In-Line
|
SPAN
|
|---|---|---|
| View all seven OSI network layers | Y |
N |
| View layer 1 and layer 2 errors (e.g. CRCs, runts) | Y |
N |
| View Gigabit auto-negotiation errors | Y |
N |
| View VLAN tagging information | Y |
N |
| Possibility of lost packets due to port oversubscription | N |
Y |
| Possibility of degraded network switch performance | N |
Y |
| View internal network switch conversations | N |
Y |
| View EtherChannel | N |
Y |
In-Line Fiber Matrix Switches
If deploying an In-Line fiber switch, there are also two other items consider - the switch's split ratio and the switch's light source.
A fiber switch's split ratio is mainly determined by the switch's receiver sensitivity, its' transmitter strength, and cabling.
Each time an optical link is switchped, the link suffers from insertion loss. With each new access port that is created on a fiber link, more light gets "switchped out" of the network. This degrades the nework signal. In other words, the more "light" a fiber network has the better. Therefore, when deploying a fiber switch, you should choose a switch with the highest split ratio that will provide the adequate amount of light to the connected monitoring device.
The fiber splitters in the switch also need to support the light source used on the network. For instance a 1000Base-LX monitoring device will transmit data using 1300-nm lasers over single-mode fiber. If splitters are used that support lower-intensity LED transmission on this link, laser light intensity could cause network performance to degrade, so you'll want to ensure you select an LX switch with compatible fiber splitters. This ensures the insertion loss that is dictated by the chosen split ratio will be accurate.