bitplumber.net

As competition increases in the metro for fiber transport, exciting new options are becoming available for “Clueful Enterprises” ™ from forward thinking non-traditional telco’s.

Fiber to the tower is driving huge buildouts in fiber networks to areas that were previously infeasible economically. That coupled with high count fiber cables (288 to 864 count) being deployed cost effectively means that fiber is not such a scarce resource.

The trend I am seeing is for Enterprises to lease a pair of dark fiber from their offices back to a datacenter within the same metro area (usually on both sides of a ring for redundancy) and then to purchase all other telecom services out of the datacenter where a competitive market exists. (i.e. IP Transit, MPLS, SIP trunks, etc…)

Once you have dark fiber on a metro path, you can “light it” with as much bandwidth as necessary and upgrade capacity over time when needed without having to re-contract in a disadvantaged negotiating position (due to remaining term length). Optics to push 1 gigabit down 20km of fiber can be purchased for as little as $7 each (you read that correctly!). 10 gigabit over 100km can be as low as $350/each. Stepping up modestly in cost, you can push 40x 10g waves over that same pair of fiber for 400 gigabit total (on each side of the ring).

One thing that makes this kind of deployment so tenable is the lack of need for specialized transport equipment. You can now put long distance (1550nm) and even DWDM optics in regular old desktop switching platforms. The only equipment needed at an office building is now basic network switches – 100% of the servers can be located in datacenters or in the cloud.

When making use of dark fiber (with diverse paths) there are actually significantly less single points of failure than if that exact same fiber was “lit” with Ethernet Private Line services from a provider. In a classic EPL scenario the CPE device is a single point of failure, the client facing optic on that CPE device is a single point of failure, and the aggregation router back at some hub site is a single point of failure. Additionally, you are dependent on power reliability at that hub site, plus depending on ring architecture, potentially dual failure scenarios with many other customer premises on either side of the ring (i.e. during a widespread power outage).

With dark fiber, no amount of typos from someone in the NOC, software upgrades, or automation gone wrong can take down your connectivity (your fiber has to actually be physically broken to make it stop working).

While many see dark fiber deals as a negative for telecom companies, I actually see many upsides:

1. Contract terms are longer – I advise enterprises to co-terminate fiber leases with their office space leases – typically 5-7 years vs. a maximum of 3yrs for lit services.
2. Deal size is typically larger as Enterprises do see the value in having effectively unlimited bandwidth that allows them to access other services in the datacenter where market economies exist. This architecture can allow the elimination of traditional “server rooms” which are large portions of many Tenant Improvement budgets and take away expensive office square footage.
3. CAPEX and OPEX for equipment is ZERO. Splicing costs may be more than for a traditional lit services deal, but that is easily offset by the longer term length and the lack of need for expensive equipment.
4. Building right of entry agreements may be easier to come by and less costly when you have no powered equipment on-site that requires 24×7 access.
5. Competition is lower when offering dark fiber solutions. In a traditional lit services transport deal there are often 5-6 carriers available, but due to fiber and policy constraints taking it to a dark fiber deal may reduce that to 2-3 options (or less!).

-Eric