The Wit & Wisdom of David Hughes

LONDON -- David Hughes has little patience with telecom service providers who make excuses or blame others -- regulators and vendors -- for their problems or pace in managing a digital transformation.

As vice president of engineering for PCCW Ltd. (NYSE: PCW; Hong Kong: 0008), Hughes is familiar with the challenges of getting legacy operations and support systems ready to handle virtualized infrastructure and demand for faster, more nimble service delivery. He's overseen the Hong Kong-based carrier's data center evolution and NFV development for the past three years. But speaking at last week's "OSS in the Era of SDN & NFV" event in London, the former US Marine said some of telecom's problems are self-inflicted.

"What is holding us back is we've got a bad case of the 'poor mes'," Hughes said "Poor me, I'm regulated. Poor me, I can't do all these things. Poor me, I don't have any money. I was in the Marine Corps, for what it's worth, and we had an expression there, we can do so much with so little for so long that we can do absolutely anything with nothing."

David Hughes, in Action

Earlier in the day, he had also challenged telecom operators to start doing more things for themselves instead of relying on and complaining about vendors.

"Maybe it's time we mature a little bit, and learn how to do for ourselves," said Hughes, whose 15 years at PCCW hasn't dulled his Southern drawl. "We are always talking about this whole need to reduce costs, and that's up there with 'the government is here to help you,' it's a lie. You are never going to reduce costs, what you hope to do is better manage your costs."

Hughes didn't just chastise the industry, however, he also had some solid advice. To wit:

  • Telecom today is "actually three businesses that run in parallel. Yes, you do have a legacy business but secondly, you need to be working on businesses that attract rapid growth and that model has different tools, different processes, different skills and you have to plan for that. And thirdly, you have to also be developing the things that might be coming in five to ten years, the bets, that are emerging. So you need to be innovating on process with your legacy current stuff, you need to be innovating on business models with the new stuff to attract growth and then, you are just looking to disrupt."

  • Breaking down linear thinking means engaging an entire operation in innovation, not confining it to one "skunkworks" group. "When the skunkworks pops up and says, 'Here I'm done,' everybody says 'We can't put that in,' so the legacy resists the change. We need to flip that model around. And the legacy needs to take ownership for absorbing the organizational debt created by these new business models and this emerging technology and feed it back in. The skunkworks approach is actually retarding growth and innovation because it pocketizes it. It doesn't help the whole company to innovate, you have one little group innovating instead of whole company innovating. We need to realize we are all in this together."

    Want to learn more about the technology and business opportunities and challenges for the cable industry in the commercial services market? Join Light Reading in New York on November 30 for the 11th annual Future of Cable Business Services event. All cable operators and other service providers get in free.

  • On managing change within PCCW and getting others to accept things such as DevOps working procedures, Hughes has a strategy. He doesn't use the industry terms like "cloud" or "DevOps" and he doesn't hit his team with the big picture all at once. "I don't say we are doing DevOps. I just said, we are going to do some automation. You lay out a crumb trail and they follow. You have to balance how much you tell them about the big picture, because if you tell them everything, they're going to quit and go work somewhere else." So rather than overwhelm his staff, he manages expectations and workloads, and doesn't imply that there is a massive transformation afoot. "People grossly overestimate how much they can get done in a year and they grossly underestimate what they can get done in ten. If you pitch it all, they get overloaded and they can't get anything done."

  • Competing in the digital services space doesn't mean going up against web-scalers, it means doing edge-computing and other strategies: "It all comes back to FOG and edge cloud architecture. You are not going to compete with OTTs because they will outspend you, they will out-nimble you and we are regulated. The best strategy for the carriers is to look more to the edge and pushing content and processing toward the edge."

  • New things in telecom used to come with new protocols and that moved slower. "In the protocol days, vendors had to wait for standards to put the new protocols into action. So we'd all sit on the river bank and watch the water go by for five, six, seven, eight years until the standard was there, it was in the hardware and we could use it to build the system that customers were asking for eight years ago. It was a wonderful thing, we could take our time and be very patient and diligent and risk-averse and all these wonderful things."

  • In today's network, things are API-driven. "When you are thinking, I can't use an API until there's a standard, that is linear telco thinking. But that's not the case. And the OTT providers who aren't used to the protocol-driven 'Let's wait eight years to get anything done,' they are jumping right on it, plowing ahead in the absence of standards. Not to dismiss standards, standards are still relevant, but maybe we need to rethink what standards need to be for."

The last comment came right before Hughes shared his thoughts about common data models, which have already appeared in Light Reading twice. If you haven't seen them, just search for "pigs will fly." (See Heavy Reading: Common Data Models Unlikely and ONAP Takes Flak as Telcos Prep for Release 1.)

Because he used humor and folksiness, Hughes took some of the sting out of his critical comments, but they were nonetheless quite pointed -- and timely. The good news is, he projects a real sense that telecom operators are far from helpless when it comes to charting their own course, and directing their own change.

— Carol Wilson, Editor-at-Large, Light Reading

Page 1 / 3   >   >>
kq4ym 11/21/2017 | 10:58:53 AM
Re: Crumbs Yes, imagine that scenario if all the transport vehicles just died. Or all the power stations in an area. Quite a lot of question and answers awaiting to solve the complex issues and tradeoffs to be considered in systems design.
mendyk 11/10/2017 | 2:12:00 PM
Re: Crumbs There are trees, and there is a forest. There's a glass of water, and there's an ocean.
brooks7 11/10/2017 | 2:08:45 PM
Re: Crumbs Actually it is an old thing the "automation of thinking".  Has been happening for about 30 years in the CAD/CAM world.  Heck I wrote an Automatic Test Program Generator in 1980.


mendyk 11/10/2017 | 9:39:12 AM
Re: Crumbs There are exceptions to every generality. For every Wozniak, there are 100 non-Wozniaks. Or maybe 117 of them. And they basically work from the same playbook. As for this iteration of automation being different, it will be when the degree of automation allows for autonomous operation of networks. That's not something that will happen for a while, but it will happen. The focus of automation -- not just nonmanual labor but also thinking -- also makes this different, but that's not specific to telecom. As for outsourcing, you get what you pay for. I wouldn't trust a mission-essential job to temps. I'm not even sure the CXO playbook would advocate for that.
brooks7 11/10/2017 | 2:18:39 AM
Re: Crumbs Dennis,

It is that last bit that I would argue with.  That is a pretty broad brush to paint say Steve Wozniak with.  Employment is a business transaction.  An employee (including the CEO) is selling their time and effort for the money they make.  Good leaders recognize good employees and work to retain them when they can.  But circumstance does not mean that this is a permanent thing.  It is like any other contract.  Unfortunately, people talk about "loyalty" to a corporation.  The purpose of a corporation is to make the shareholders money.  Any notions beyond that are just invalid.  Why would some shareholder in some mutual fund care anything about any person at the company?  All of that is lies that we like to tell ourselves.  The best public venue for that is the NFL.  The day that a player is no longer useful, he is cut.  If there is public sentiment for him (which might lead to lowered revenue if he is upset), then the team might ease him out of his job.  I assume that is why they sign indivdual contracts, otherwise teams would have to go through the entire process that normal employers do to fire someone.

And you have posted about automation and job loss in other threads.  In at least one of those, you considered this automation different than other examples.

And there is a completely simple way to deal with retention issues.  Outsource the work if people quit so that you have a better flexible staffing model.  Since none of the work that they do does not matter in the long term, the loss of the IP to the subcontractor really doesn't matter.  Which, by the way, is one of the huge advantages that the cable guys have had.  


Duh! 11/9/2017 | 2:56:44 PM
Re: Crumbs As long as we have TCP slow-start and congestion avoidance, latency is a proxy for throughput. In an era of 1G access/100+G transport, prop delay is the overwhelmingly dominant source of latency: 4.9 µs/km.
mendyk 11/9/2017 | 9:39:45 AM
Re: Crumbs Not sure where you get the "this is different" message. In this specific case, you are pulling in a limited number of people who are going to enable the automation that will let your company ultimately reduce its headcount in a significant way (theoretically). If your concern is that you won't be able to retain the people who will make that transition possible, then you need to figure out a way to retain those people. It's pretty simple -- and has almost no chance of being adopted because people in management tend to have close to zero respect for the people whose work makes their own jobs possible. And they tend to have very limited abilities to solve problems creatively.
brooks7 11/9/2017 | 1:34:07 AM
Re: Crumbs That is an optimization Carol.  They better be able to drive when the network is completely dead or they will go noplace - literally.  So they might be connected, but completely 100% functional disconnected.  Imagnine a DDOS attack stopping all cars in the US.  See what I mean?


Carol Wilson 11/8/2017 | 11:23:46 PM
Re: Crumbs Self-driving cars  will also be connected cars. Traffic and safety information will still have to come into the car, from external sources.  
brooks7 11/8/2017 | 7:45:32 PM
Re: Crumbs Of course, if you put the computer in the car then you don't have to worry about the hiccup in comms driving your car into the wall (in other words, self driving cars MUST be self-contained).


Page 1 / 3   >   >>
Sign In