Top 10 considerations for building an embedded xWDM network
Thinking about rolling up your sleeves and designing a CWDM or DWDM network? You probably understand the basics of wavelength division multiplexing and its great potential to provide capacity expansion and flexibility at a cost you can afford. Now you need to know where to start.
The first step in the network design process is assessing both your current requirements and the need for additional bandwidth as your network grows. Next, there are some important factors to consider before you can design and build the network. Here’s a step-by-step guide, with practical considerations and recommendations, to help you get the best fiber optic network for your specific needs.
1. Dark fiber availability
To build an embedded CWDM or DWDM network, you’ve got to have access to dark fiber. Without it, your only option is to instead lease a service from an operator.
2. Fiber pair or single fiber?
Once optical fiber access is confirmed, you need to know whether you’ll have access to a fiber pair or a single fiber strand. This affects the choice of components and also the capacity of the network. Remember, a fiber pair can handle twice as many channels as a single fiber strand. Both CWDM and DWDM networks can be built with a single fiber or fiber pair.
3. Know what loss to expect as well as the distance of the fiber route
Many people building an embedded xWDM network will use the distance specification of the transceiver and the distance of the fiber as the key criteria for designing the network. But this is risky. Under ideal circumstances, a DWDM and a CDMW ZR transceiver can theoretically span 80km. Yet this distance specification of an optical transceiver is only a guide. In reality, disturbances usually occur along the way from the transmitting to the receiving end of the fiber, such as mux/demux losses, fiber losses and patch losses. To ensure that a network can operate fit for purpose, ensure that the OTDR results are known that give the exact losses of the fiber.
When it comes to designing an embedded network, if the network losses and distances fall within the scope of what can be achieved by the optical transceiver (CWDM 14dB ER / 23dB ZR and DWDM 14dB ER / 24dB ZR), then a purely passive network is possible. If the system losses are higher than this, amplifiers will be required.
4. Network topology and number of sites
Are you building a point-to-point network between two sites? Or a network that connects multiple sites perhaps in a campus-style network? Do you need a resilient network with a north and south route in case of a fiber cut? These are all questions that will come up when you order the dark fiber, but they also affect the component choice when designing the network. If connections are required at interim sites, OADM modules (optical add/drop multiplexers) may also be considered, which bring with them additional losses that you should take into account when making the design.
If a resilient network is required, this usually means two fiber routes; one in the “north” route and one on the “south” route. If a fiber cut happens in one of the routes, then the network then switches across to the other route. This resilient approach is known as a “working” and “protected” route.
When building a “passive network” i.e. a network which uses WDM Mux/Demuxes and OADMs without power, then the optical power budget of the network is essential. We covered Power Budget and Optical Link budget in one of our previous articles but the basic premise is that the power budget of the transceivers must be enough that it allows the signal to get from site A to site B and through any losses that may arise in any mid-point OADM sites.
5. Is monitoring required?
Typically, embedded networks are passive and therefore have no ability to offer any type of signal monitoring. Smartoptics DCP-M open line systems, for example, have multiplexers with monitoring and distance extension circuitry built in. Otherwise, an embedded system is not possible, and active systems using additional transponders and management systems are required instead.
6. What are you connecting?
Be aware of the data rates and transceiver form factors. Ethernet switches, Fibre Channel switches and Sonet/SDH equipment all use different protocols, so different transceiver types need to be used. Different data rates also potentially require different form factors (SFP, SFP+, QSFP, QSFP-28, QSFP-56, CFP/CFP2). And not all transceiver types have xWDM variants.
7. When to use compatible transceivers
Not all switch manufacturers offer WDM transceivers. And even if they did, you might choose not to buy them. In either case, you can use a third-party transceiver – often called a compatible transceiver – to make the connection. The challenge is that this might cause problems if something goes wrong in the networking setup. Switch vendors tend to limit the support to solutions involving non-verified components, making it a risky option. Or else, have in mind that you can use whatever transponder system you prefer to convert the SR/LR signal from the switch to a CWDM or DWDM signal that can be used with your transmission network.
8. Number of channels for current and future needs
Think about how many channels you need to connect. 1? 5? 20? 50? This affects the capacity of the multiplexer you need to use. But you’ve also got to consider how much headroom you want to leave for future growth. If you require 4 channels now and a maximum 6 channels in the future, then an 8-channel mux/demux makes a perfect fit, with the lowest cost and enough for today’s and tomorrow’s traffic growth. If you need 16 channels now with another 16 over the coming 18 months, then a 32- or 40-channel variant is better.
9. CWDM or DWDM?
A common misconception is that one is cheap and one is expensive. This is not the case – both technologies can be used in an embedded environment (xWDM transceiver into a data switch and connecting to a multiplexer). And both can be used with transponders (media converter that converts SR/LR signals to CWDM or DWDM). A better comparison is that CWDM is limited to 16 channels (or 8 for longer distances) and that it cannot be amplified. DWDM typically offers 40 or p to 80 channels and can be amplified, and is better suited to higher order signals such as 100G and beyond.
As a general rule of thumb, if you’re starting from scratch with a green field network then DWDM is generally recommended over CWDM. It can handle more channels and higher data rates than a CWDM solution and can be amplified. If used in embedded networking, the cost of both technologies is comparable. For more detailed information about these two alternatives, read Coarse and Dense Wavelength Division Multiplexing.
10. Encrypting the signals
If a signal needs to be encrypted over an optical network, a transponder system, with electrical elements is required to provide the encryption. The DCP-1610, for example, is a high capacity multichannel transponder allowing up to 20 channels of 1-16G traffic to be connected per 1U of rack space. Encryption cannot be built with an embedded passive WDM network.