I realise I've "taken my eye off the ball" a bit over the last few months with regard to operators' strategies for upgrading their ageing, low-bandwidth backhaul networks. I'd sort of assumed it was in the progress of being fixed, and was only a short-term issue.
It's not as simple as that.
I went to an event last week called NetEvents. I've been going for years - it's an interesting 2 days of presentations and a dozen or so pre-organised vendor briefings, speed-dating style. (The bell rings and... "next!"). Mobile/wireless is only a small part, so I'm forced to catch up with bits of the networking industry I don't normally get exposure to. This time, there were a variety of metro ethernet vendors, as well as the metro ethernet forum, and some people doing clever things making multiple copper pairs work together where fibre isn't available. (step forward Hatteras, Actelis, Resolute)
I'll be honest, I find most in-the-background things to do with metropolitan fibre or copper networks deathly boring. But like many dull-but-worthy areas of the tech industry (I used to analyse enterprise storage for a while too...) it has a story to tell. I tackled the vendors about how fast they were seeing the much-ballyhooed wireless backhaul upgrades going.
The general consensus seemed to be that in the US and bits of Asia, things are going pretty well. Most North American cell sites have decent upstream capacity. But in Europe and some other countries, especially where there are older 3G networks (and limited fibre installed), progress remains very slow. Yes, this varies a lot between operators, and sure there's also a lot of microwave installation going on. Metro ethernet and all that cool stuff is still "in trials" rather than being rolled out aggressively in many wireless networks.
But the bottom line is that there's still an awful lot of base stations out there with 2 or 3 E1 connections (ie 4 or 6MBit/s connection to the network). Some of these are old and can't even support ethernet/fibre if the operator wanted, and will need outright replacement instead. These are bottlenecks as we move towards HSPA, LTE and so on. There's no point putting 100MBit/s on the radio side at a cell site if it's connected back to the core with a piece of wet string.
One of the problems is that the existing networks aren't at full capacity, and so convincing CFOs to stump up the case to replace them is tricky.
The paradox is that the "headroom" is now much lower. Imagine you have a 4Mb pipe from the cell site, and are selling normal WCDMA at 384kbit/s data speeds. You monitor usage... and at peak times maybe it's getting 3 users... then 4 users the next month.... 5 users two months later... you get time to plan an upgrade so that capacity is in place by the time 10x384 = about 4Mbit/s. Now move to today, launching 3.6Mbit/s-capable HSDPA devices. You monitor usage again, and this month you have zero users... then 1 user next month... and it's full.
Let's see how long it takes before the Metro ethernet community really reports some movement on actual deployments, representing a decent proportion of the installed base of cell locations.