The sun beats down from a clear blue sky every day. The noonday temperature is never less than 30° C. Sand everywhere, with the warm blue sea a refuge when it’s all just too hot. This is the upside to working in the Gulf, where I was on contract for the first part of this year.
Building a new network from scratch is always a delight. The vendors are keen to showcase their hottest new technologies, and we get to think very seriously about what our customers will be doing with telecoms over the next five to ten years. I had been commissioned to look at a carrier metro network surrounding a new airport complex in the UAE and we had our share of issues.
1. The access network
The client quickly agreed that we needed an MPLS core to concentrate the traffic and hand it off to the national licensed operators. But how to connect the core to residential and business customers? Our first thought was GPON. We could feed 32 or 64 user terminals from a central office GPON card at 2.5 Gbps downstream and half that upstream. Seemed plenty of bandwidth for the next five years.
Then we started to think about IPTV and VoD. This is traffic which is both high-bandwidth and incapable of statistical multiplexing. If you assume 8 Mbps for a high-definition IPTV channel, a few TVs per household and you’re starting to talk real bandwidth. Perhaps an active Ethernet solution delivering 100 Mbps dedicated to each household was more future proof?
We modelled and we costed. The debate went one way and then the other. GPON is more cost-effective than an equivalent active Ethernet deployment in most geographies. Provided you stay within the GPON bandwidth envelope.
The client is still thinking about it.
2. WiFi vs. WiMAX vs. 3G/4G cellular
The client’s first thought was WiFi everywhere. It’s the modern thing, isn’t it? And we want to connect to the Internet everywhere. And with security being what it is, we want good surveillance and communications. After some modelling, and a budgetary estimate, we started thinking instead about WiMAX.
WiMAX can be thought of as a higher-powered version of WiFi. We could cover the metro network area with WiMAX at a fraction of the cost. But will it take off or is it the Betamax of wireless networking? If we hold off the decision for now, when will it be safe to take it?
A complication for the client was that the cellular guys were going to cover the area with 3G base stations anyway (HSPA => LTE). Functionally this is pretty much equivalent to WiMAX. Maybe the right answer was to be a virtual network operator on this infrastructure (or just buy bulk capacity)?
There were opinion-formers within the client who were attracted to the idea of pervasive surveillance. If you want to cover hundreds of square kilometres with 24/7 real-time video surveillance, that’s a lot of ~4 Mbps video streams to carry across the network to your video head-end. It’s not impossible if you have the budget, but the more interesting problem is what to do with the feeds in that hangar-like control room. We’re talking thousands of camera feeds, almost all of which show absolutely nothing of interest.
It wasn’t in our brief, but I wouldn’t be surprised if there were contracts at some stage for some really state-of-the-art automatic image-processing and threat analysis computing platforms.
4. IMS and new services
Publicly we are all in favour of IMS. Privately we ask ourselves: what are the new services, will any of them ever make sufficient margin and will they be destroyed by cheap or free offerings across the Internet a la Skype?
So this was the debate we had with the client. Do you want to be a mini-clone of an established telco, developing a vertically-integrated services stack? Or might it be better to create platforms for general service innovation (OK, maybe even IMS with a service development & delivery platform) which you can open to the world and charge rental for?
The latter resembles that tried and true business strategy of running guns to all sides in the conflict. Always where the real money is made.
You always learn a lot from designing a state-of-the-art network. We had, for example, a huge discussion about the use of VLANs. This is industry ‘work-in-progress’ in trying to make Ethernet do the kind of per-user, per-service virtual circuit which was so easy with ATM. Perhaps the engineering can be made to work once some of the newer standards are bedded in, but it’s undeniably kludgy at the moment.
I could talk about some of the other issues. How much does the client need the whole panoply of the next-generation OSS (NGOSS)? What’s the role of DWDM and an optical transport network when you built-in plenty of fibre? If the metro traffic core is going to hit the Terabit per second level in the next 5-10 years, can this be done scalably or do we need to rethink the underlying hardware and protocol architecture?
But enough already.
© Interweave Consulting, September 2008.
-- PDF version of this newsletter here.