An Interview with John Hayduk

Editor’s Note: There is no doubt that the telecom industry has experienced a lot of growth over this past year – and a lot of challenges. Carriers and service providers have had to build out their networks and increase capacity to handle the extreme surges in bandwidth usage.  There is always more than one way to address these challenges and it is important to fully consider each approach and model.

SubCableWorld recently had the pleasure to speak with John Hayduk, most recently the Chief Operations Officer at Tata Communications whose career began at Bell Communications Research as a Senior Engineer in 1990.  He is a strategic partner for NJFX, the only Cable Landing Station (CLS) colocation campus in the U.S offering Tier 3, carrier-neutral data center capabilities.  John gave us his thoughts on these challenges.

John Hayduk: When I see the things that are happening in the industry, I get concerned especially about some of the companies that say, “Hey, bring your subsea cable glass through a certain amount of terrestrial land to where my infrastructure is and that will be a good platform for you to draw on.” 

There are two things to consider here: 1] the characteristics of the subsea cable, installation and failure rate and 2] the characteristics of a terrestrial network.  Trying to combine those two, at least in my opinion, isn’t something that I’d ever want to do.  From a network design perspective, you’re bringing the failure characteristics of the terrestrial network onto what is basically the tail of your subsea network. 

Subsea technology has improved significantly over the past 10-15 years.  Increases in the number of fiber pairs and the amount of data that those fiber pairs can carry have massively increased.  Now it’s well over 100 Tbps in some cases.  Submarine cables are still expensive to deploy and to put all that in, and link it to a 20 or 30 or 50 miles terrestrial tail just isn’t the best solution.  The break/fix characteristics of a terrestrial network are way worse than a subsea network. 

It all comes down to frequency of failures.  Whether you bury a cable or run it overhead, you have people digging, expanding highways, etc.  The failure rate is an order of magnitude greater for a terrestrial network versus the subsea segments.  Even in your best-case scenario, you’re still talking about hours of downtime.  They call the tech in, figure out where it the problem is, what crew to send out, are all the people available and how quickly can they get out to the right places.  Now you’ve just stranded that asset. 

Having a true cable landing station and data center campus gives you the best of both worlds.  You bring the submarine cables directly into a safe facility and then having a choice of many different carriers out of that facility.  It will let you design whatever redundancy you need in the most effective manner for your network.  You’ve taken advantage of the dependability that the subsea cable network provides you, because it’s very infrequent that they fail.  It takes a while to fix them but in the end the number of failures on subsea cables compared to the number of failures you’re going to have on a similar length terrestrial network span is massively different. 

You’re going to take hit after hit after hit on the terrestrial side.  And that’s not all just the break and fix, because you run into a lot of carriers using the same conduit structure.  If you’re installing a new cable and the cable is running fine, chances are somebody is in there either in the same conduit or an adjacent conduit.  You may not drop the link but you’ll definitely see error rates on that link that will cause your router sometimes to reconverge.  Or there will be some transient failures for a period of time.  You’re taking a subsea cable that is designed and built to live on the bottom of the seafloor for 25 years in a reasonably undisturbed manner and you’re bringing terrestrial failure characteristics and disturbance characteristics to it. 

thumbnail IMG 8227

If I was given a choice on building a new network, that is not how I would look at putting it in.  Say you’re putting in a new subsea cable.  Can you get fiber of the same quality on the terrestrial side?  Will it all be underground?  Will some be above ground on poles where it can be hit by a storm?  You’re bringing a much more reliable, steady network architecture down to a really poor failure-type scenario, or you’re capping capacity because you can’t get the same quality of fiber. 

This is why I think the model NJFX has created is better suited.  You want to have the most efficient design and that would be to build the data center where the cable landing station is and then have the option of multiple carriers out of that because you’ll be able to design what you need for your business.  If your business wants something really thin and cheap with single redundancy, you can easily do that.  If your business wants to be a lot more resilient and you want to put in two or three levels of redundancy in, you can do that as well, ideally with multiple subsea cables and multiple terrestrial networks.  You can build in whatever level of redundancy that’s affordable for your business.  And that’s the best solution, versus making a big compromise and bringing a much bigger failure rate or disturbance rate on top of a more stable network.  That’s not really how things should be designed, in my opinion.