Thinking small

We spend a fair amount of our time talking with access equipment vendors about their newest product lines and how they market them to the carriers. So we hear a lot about these vendors’ strategies and messaging – for better and for worse.

We spend a fair amount of our time talking with access equipment vendors about their newest product lines and how they market them to the carriers.  So we hear a lot about these vendors’ strategies and messaging – for better and for worse.

Scalability has always been a catchword for telcos, because they like to know there is some futureproofing built into the product.  Indeed, some products need to scale big and fast – like BRAS, routers, switches and the like – because these are things in the network that should get bigger, denser and higher in capacity.

However, what seems to have happened lately is that in an effort to see carrier class, scalability is being touted on all sorts of devices in the network where we wonder if that’s one of the top three things that should be being pitched. It makes us stop and wonder what is the right size for specific devices, given where they fit.  For instance, we have to stop and ponder when we hear a vendor pitching their access aggregation platform as supporting X thousands of lines.

Don’t get us wrong – our penchant for pausing is not to say there isn't a market for big CO-based DSLAMs and the like, and that there aren’t spots in the market where highly scaled access line aggregation comes in handy.

What we’re questioning is whether that message is still the best message to be so high on the ladder of benefits being pitched to carriers in light of the major trends in applications and services leading the way to higher bandwidths on a per customer basis. Scaling in terms of lines is one thing, but when each line itself will need ,much higher bandwidth to support IPTV and other bandwidth hogging apps. (We promise not to use the term “triple play” in the column today, but feel free to read that into what we’re saying here.)

And unless someone redefines the laws of physics, all  of the high bandwidth access technologies that are being considered in the near- to mid-term are built around systems that either require short loops (DSL technologies) or small subdivisions of users (PON ).  On the DSL side, the few thousand foot loop lengths required for 20 or Mbps per user take you down to the neighborhood level for the first round of access concentration. For PON, dividing up even Gigabit speed fiber between more than 32 users isn’t really an option.

Either way, you end up with access devices that are serving users in the tens, not the thousands. 

A similar conclusion can be reached on the backhaul side of things – even as more vendors tout their Gig-E backhaul abilities, the number of users who can be supported with, say, Unicast (or minimally multicasted) HDTV streams is a finite number and it’s not in the thousands.

So what is the right number? What size access concentrators should vendors be developing for telcos? Well obviously, one size doesn’t fit all, and as we said earlier, there will still be situations where a big multi-shelf hundreds or thousands of line DSLAM still fits into the network. But for most customers, especially in suburban and rural areas, there’s a sweet spot somewhere under a hundred.

So we called around and found a lot of people agree.  For instance, the folks at Pedestal Networks tell us that their research points to 48 lines as an obvious place to aim. Of course Pedestal is in the business of selling small DSLAMs, but we think that their logic makes sense.

Pedestal also did some research on customer networks, and put together some analysis that correlates loop distance, cable size (number of pairs) and number of customers.  They also did some analysis on the real throughput provided by a GigE connection, and the actual requirements of customers using video services. Using this real world data combined with some conservative bandwidth engineering, they found that even in urban locations, there were obvious clusters of 30 to 60 customers who might be described as a VSA or video serving area. 

You can do your own math – looking at your own assumptions about video codec adoption, multicast, and backhaul oversubscription. Run your own numbers. We can’t predict exactly what you’ll find, but we are willing to bet that you’ll find something in the same ballpark.

But please, if you’re pitching us solutions for carriers, let’s make sure the numbers fit the applications.

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.

Copyright © 2005 IDG Communications, Inc.

IT Salary Survey: The results are in