Loft44 Group

IP and Ethernet in the telecom space really disrupts the traditional eco-system of products which make up a network system.  When I was in the access space, I could understand why customers with limited numbers of fibers in the ground would use a transport specific product and some form of WDM, wave division multiplexing, to avoid burying more fiber.  What I couldn’t quite understand was the value proposition for a transport only platform that implemented layer 2 and layer 3 transport level protocols.  These protocols have been implemented and in use by the transport gig–e and 10 gig-e ports on access equipment, switches and routers for some time.  After studying the space for the last couple of months, I’ve reached the conclusion that there exists the need for different levels of middle mile equipment sophistication and that carriers should follow a few guidelines for selection.

Following the KISS and KII (Keep It Inexpensive, pronounced “key”) approaches, there are several approaches:

(1) PASSIVE MIDDLE MILE: Implement middle mile by leveraging the transport which exists in your Access equipment by using a passive middle mile layer for MUX/OADM and optical switching.

(2) ACTIVE MIDDLE MILE:  Implement middle mile by using an active platform that consolidates transport protocols and optical MUX/OADM/Switching

The passive approach is an order of magnitude less expensive than the active approach when you take into consideration the total cost of ownership to include capital equipment, installation/integration,  ongoing maintenance/repair, and ongoing management.  Passive approaches are best suited for optical needs which are anywhere from 2X-60X  the bandwidth you are currently using in a “grey” fiber application (a single channel).  The upfront planning and design is limited because the network is so simple.  If future requirements are capped at up to 18X the bandwidth, the less expensive CWDM route can be used.  The middle mile becomes a “set it and forget it” implementation assuming you use a passive system with proven high quality and low insertion loss.  The only potential drawback is that if you use DWDM, adding channels in the future can require optical power rebalancing.  Otherwise, this approach is simple, straightforward, and kind of “fits” with the traditional telco deployment strategy that separates passive “cross connect” frames from active electronics.  Your active electronics, say the routers and access system, would employ CWDM SFP/XFP or DWDM SFP/XFP and you’d implement transport resliency/reconvergence ONCE in these platforms. So protocols like EAPS, MEF, EPS, RST, SONET etc would be implemented just as if the underlying network was a single strand “grey” optical network.  Yes, this network can carry Ethernet (Gig-e, 10Gig-e), SONET and TDM on the same fiber!

The active approach becomes more applicable when your access equipment and/or routers do not implement resiliency protocols that guarantee an appropriate reconvergence/recovery should an issue occur.  It also become more applicable when you need to convert SONET to Ethernet (for example), so that an optical->electronic->optical conversion is necessary. In these cases, your access/routing equipment links would use grey optics (single channel 1310nm) and implement a minimal transport protocol — but be careful you don’t duplicate transport on top of transport in this case. It should work, but why impose this level of complexity? It’ll just cause operational problems you’d be best to avoid. These platforms will typically accomodate SONET, Gig-e, 10Gig-e, and even TDM. Note that you’ll end up purchasing 3 optical “lasers” compared for every 1 optical laser you need in the Passive approach (two to connect the access/router to the transport in the form of SFPs, and one to drive the optical middle mile span).   As we progress, these platforms seem the most likely place to implement a ROADM, which is essentially more logic and electronics to allow for an on-the-fly reconfigurable optical layer.  Again, using KISS and KII, although there will be applications for ROADM, the mainstream optical network that is effectively planned should not require this level of sophistication.

Comments?