Start thinking about buying WAN bandwidth by the gigabit. That's the advice of Desh Deshpande, CEO of US-based Sycamore Networks, which is building optical gear for carriers. Following his success as founder of frame relay pioneer Cascade Communications, Deshpande has turned his eyes to the next frontier: optical networking. He says new optical hardware will make bandwidth so plentiful and so easy to set up and manage that enterprise network executives will stop thinking about the WAN as a place in which bandwidth must be conserved. Deshpande recently talked with IDG's Tim Greene about the future of optical networking.
IDG: What impact will optical networking have on enterprises?
Deshpande: First, ISPs will start getting better backbones and therefore the service customers get through ISPs will improve.
Second, it's pretty clear that all MIS architectures are moving toward intranet- and Web-based applications. Also, everybody is going toward virtual private networks. VPNs will catch fire really quickly, but then people will find that T-1 and even T-3 is not enough bandwidth to run some of the applications, especially when you start interconnecting the headquarters and the main sites of a large company.
When people start running out of that bandwidth, what they will need is OC-3, OC-12 and OC-48 type of connections. At those speeds, it doesn't make sense to tunnel through the Internet. Carriers will offer optical VPNs.
What's an optical VPN?
You use the lightpaths to offer high-speed services to corporations. These lightpaths will carry the IP traffic, but the IP traffic stays within the company.
What would a corporation do with an OC-3, let alone an OC-48?
An OC-3 is not that much. It's 155Mbps. If you look at a lot of major sites, the backbones are all gigabit - in fact multiple gigabits - within the campus.
That's why you're seeing these multi-gigabit hubs all over the place. If you have a thousand people at one site and a thousand people at another site, and you have collaborative work going on, 155 megabits to interconnect those things is not unreasonable.
Won't that be mighty expensive?
Prices will come down 20 to 30 per cent per year - forever.
What I think will happen is that OC-48 is kind of a sweet spot in the sense that it's 2.5Gbps, which is an enormous amount of bandwidth by today's standards.
But it's a pretty easy technology to work in optics. It's like the 56Kbps modems' kind of technology. OC-48 will become like T-1s - they'll be everywhere in about five years.
IT executives in large corporations will have to think about this when?
I think they'll start thinking about this in six months, and you'll see installations in less than 12 months.
What comes next?
Within a couple of years, they will be able to signal to the optical network and come up with new applications. Within three or four years, the carriers will start opening up their networks and will be saying: "Not only can I get you a lightpath, I can direct it wherever you want it to go."
So if you have a network that connects to Boston, Chicago, New York and Dallas, you can tell the network where you want the bandwidth and for how long.
It's not a technology enterprise staffs will have to understand intimately, is it?
Right. What they need to understand is the capability of that technology so they can go and demand it.