Broadband Cumbria

Part Two of Barry Forde‘s smash-hit three part series on how the communities of the Eden Valley could get Fibre To The Premises by themselves:

Step 1. We would start with a map of Eden District with the 74 Parish boundaries plus the 4 Penrith Wards marked on it. Look at each of the 78 areas and identify a logical central point within each where FTTP hub equipment could be located. For instance, many parishes will have a village or cluster of properties within them, and this would be the logical central place to locate a network FTTP hub.

But it would be wrong to assume that this place would be in the geographical centre as in most instances it would not be. For very small parishes, it might be more logical to attach the properties to an adjacent parish to avoid having too many very small hubs.

It would probably be sensible to set the cut off point at around the 100 property mark, which would take 28 parishes out of the list, leaving 50 FTTP nodes across the district. But this would be up for discussion as the design matures.

Step 2. would be to talk with parish champions to see if they consider it a good location from which to run fibre out to reach all properties within the parish. If so, could they then identify a suitable property to attach the hub to? For instance, might there be a village hall with spare space? Or could a a small annex be built on to house the kit?

If there is no suitable space, then we would need to install a street cabinet instead, and again the best location for it would need to be identified. Space requirements would vary depending on the number of properties being served from the node. I would allow one rack for equipment and then another one per 500 odd properties served.

Step 3. We would also need to identify suitable locations in Penrith and Kirkby Stephen to be the north and south end core nodes for the network. These would need to have additional space as there would be extra equipment located in them. More on this later.

Step 4. Once we have sorted out the location of the 50 nodes, we would then need to sit down with the map from Step 1 and design the interconnect. Starting at our core node in Penrith, we would draw a straight line to the nearest node towards the east side of the valley.

We would then carry on drawing lines in a daisy chain linking all the nodes down the east side of the valley until we reached the bottom at Kirkby Stephen where we end at the southern core node.

Then, we would repeat the exercise coming out of Penrith and going down the west side of the valley linking all the hubs and ending at Kirkby Stephen again. We would now have a ring round the district, a very jagged ring, but logically you can see it is a ring.

Step 5. The ring is the network backbone which would connect each FTTP hub to the core nodes in Penrith and Kirkby Stephen. We would then install a 144 fibre cable in the duct connecting all the nodes and allocate 2 fibres for each node. So node 1 would get fibres 1 and 2, node 2 would get fibres 3 and 4 and so on. Assuming we were to go with 50 nodes, and they were split equally on the east and west routes, we would need 25×2=50 fibres. However, we would be wise to install 144, so that we would have plenty of spare capacity for expansion when/if needed.

Step 6. Each hub would have a network node installed, which would have two 10GbE ports plus a number of 1GbE ports ready for connecting properties to it. Initially, only the minimum number of 1GbE ports would be installed, but these would be added to as and when properties were connected.

At the two core nodes, we would install a large switch router, such as a Juniper EX8216 with, say, 64 off 10GbE ports installed and BGP software (which has capacity for 128x10GbE ports). We would then use the pair of fibres for each hub to provide a 10GbE link north to one switch and south to the other. Under normal conditions, each FTTP hub would pass traffic directly to Penrith over its northbound 10GbE link. However, were the link go down, it would automatically reroute traffic southbound to Kirkby Stephen.

The north and south core nodes would be interconnected with fibre on both routes, so traffic would continue to flow with no visible outage to users on the route where the break occurred.

Step 7. At the two core nodes we would also install additional equipment,such as DNS and DHCP servers to provide address allocation and name lookups. We would also put in management and network monitoring kit.

For telephony, we would install a VoIP switch capable of handling 23K lines and with an appropriate number of ISDN30 circuits for local breakout. Having a pair across the two nodes would give us full redundancy and would cater for a switch failure.

We could also install a multicast head end so that video services being injected within the valley - local content - could connect to it and then be available to all connected properties. We could also provide space for Video on Demand servers and other specialised kit which service providers might want to install.

Step 8. From Penrith, we would lease fibre to Telecity in Manchester and light several 10GbE wavelengths to connect to the EDGE-IX peering service, a tier 1 IP transit provider, and other high volume data sources such as Google and YouView.

We would now have a highly resilient ring network with all nodes dual connected to high performance edge routers. There would be no single points of failure in the network, nor with any services. This would be a full carrier class network,equal to anything deployed worldwide, but of much higher capacity than anything commercially available in the UK.

The fibre routes

But I hear you saying: “hang on a minute, drawing straight lines between nodes to form the ring is impractical, and you can’t dig such a route economically, so the above might look elegant but it can’t be done”. Quite right, so we would now begin a process of morphing the backbone ring into the FTTP digs to premises.

Step 9. We would take a higher resolution map of each parish, with the node and incoming and outgoing lines shown on it. This map would need to be of sufficient scale to show individual properties clearly. A1:10K would be about right. We would now take the incoming route and look along the segment of the parish it traverses and locate all the properties along that route, both those close in and those slightly further out.

Step 10. We would then say: “OK if we wanted to run fibre to all those properties, how would we design the route so that it started at the network node and ended on the parish boundary in the approximate place that the straight line did? This new route would be a meandering animal, which has splice bullets located along its length in all the right places from which to out dig spurs to reach the properties. We would now have route 1 of the parish FTTP rollout.

Step 11. We would repeat the exercise on the outgoing straight line, and again end up with a meandering route serving a number of properties and ending at the opposite boundary of the parish. This is route 2 of the parish rollout.

Step 12. Later, we would need to complete the work by designing further routes radiating out from the hub to reach 100% of the properties in the parish. Each route would have one or more 40mm ducts installed which would contain one or more 96 or 144 fibre cables. They could be traditional fibre cables or they could be blown fibre. The decision would be made later and no doubt it would be a mix of both. At each splice bullet, we would allow for 4 fibres to be dropped from the trunk bundle for each property served from that splice point.

Then, when the property owner asks for service, we would arrange for a 7mm duct to be laid to the property and blow a 4 fibre cable into it which would be spliced onto the 4 fibres reserved for the property in the trunk cable. At the other end of the trunk, all the fibres would be presented on connectors in the second rack, and we would then use a patch cable to link two of them to the active equipment in the FTTP hub.

On routes 1 and 2, which would be the routes that also map to the backhaul ring, we would simply install an additional 40mm duct for the 144fibre cable used to provide the ring. In effect, we would have removed the labour component of digging the backhauls and end up with only the cost of the additional 40mm duct plus the fibre to go in it.

It’s about 40Km from Penrith to Kirkby Stephen as the crow flies, so allowing for the meandering routes, let’s suppose 60Km on each of the east and west routes. 120K of 40mm duct with a 144 fibre cable would cost around £3.5/m or £420K all in. Note that this would be a one off cost. Using the traditional model of digging along roads would incur a cost of around £40/m or nearly £5M.

Or to put it another way, based on 23000 premises, our backbone cost would amount to ~£18 per property on top of the cost of digging out to that property, and my estimate of £1K per property would allow for this element. So the backhaul would virtually disappear. Neat or what?

Rollout

Once the process has begun, it would be necessary to devolve a lot of planning and coordination to parish champions. From everything I’ve said above, it is obvious that they would have a key role to play and would be the decision makers for their parish.

The only external rule that we would have to apply is that they would HAVE TO dig routes 1 and 2 before any other work is done, so that we could establish the ring and provide their backhaul. Beyond those two routes, it comes down to priorities and each community would need to set their own.

It would also be down to each parish to negotiate with the local landowners to get agreement on the best routes. The exact path followed by each route would go through a number of iterations as local knowledge would kick in and farmers would tell us that there are problems with a particular field or that this field boundary would be better followed than that one due to buried rock etc. Also, the local community would need to get wayleaves agreed, and yet again revisit the routes if it these agreements weren’t forthcoming.

However, as this would all be handled at a very local level by members of the community and would be being done for the community, I think it could be done fairly smoothly.

For a complete list of parishes (and wards in Penrith) in Eden District, check this document Parishes/Wards, Eden District.