Imagine you had the choice between watching your favorite band at the local bar, or seeing a good ‘cover’ band playing songs of that very band at a concert hall nearby… what would you do? The decision looks obvious, doesn’t it? Why isn’t everything in life so simple?
Since telecommunication has grown beyond analog, two- and four-wire communication, there’s been a divide between packet and TDM. Between supposedly ‘infinite flexibility’ and allegedly ‘unbeatable reliability’. Since the so called “packet teams” were (well, are) usually closer to the money making services of a service provider, they often were in the lead. They also took the lead because the admittedly ‘more elastic solutions’ under their responsibility appear more sexy than their ‘reliable-but-not-so-dynamic’ brethren in the neighboring department. While the TDM (meaning optical) guys were fighting their way up the value chain by inventing P-OTS and adding all sorts of switching blades to their multiplexers, the packet guys kept on adding protocols to solve issues created by other protocols, and waited for the day when wide scale ‘IP over DWDM’ deployments would become the norm.
So far so good… if only there wouldn’t have been the underlying suspicion that packet solutions, while really good at delivering everything flexible, are not the ‘way to go’ when it comes to really wanting determinism and reliability. Here, an optical network is the solution ‘in vogue’, and all flavors of Fast Reroute, Resource Reservation and ‘what have you’ couldn’t really touch that. And there are many flavors, all trying to achieve the same goal: one of making a packet network behave more like a TDM network. Ouch, that hurts…
It must have been a new experience for network planners, when a while ago, under the auspice of both progress and cost competitiveness, even representatives of solid Tier 1 PTTs started talking about lowering their quality of service level agreements (SLAs), at least for their consumer customers. This was triggered, at least in part, by the aim of phasing out old circuit switches, and the realization that their IP-based successors sometimes find it challenging to deliver on the same SLAs. But the debate did not last for very long, possibly also because we saw the entrance of the new darling of telecommunications: 5G. And 5G triggered debates both about the ultimate way of delivering Ultra Reliable Low Latency Services (URLLC) and the need for ‘Determinism’. Network Slicing is supposed to be the best way of providing it.
Connoisseurs ‘in the know’ guessed it immediately… it’s time for the next round of the packet vs TDM battle!
At a customer meeting the author of this lines attended, we were accused of “adding TDM to a packet network”. It was as if we had proposed to sell the furniture of their meeting room on a flea market… just because we mentioned that: “FlexEthernet, in some parts of the network, has real values’. But otherwise this meeting was friendly and peaceful, also because we were praising concepts which to some might look like creating little TDM links inside a router to keep both latency and jitter down. Nice to see that also routers have now turned into P-OTS…
Listing all different concepts that fit under the TDM vs packet heading and their pros and cons would turn this little blog into a novel, but it’s fair to say that once again, managing those flexible packets in rather rigid ways is the underlying idea. And if one would delete the term Segment Routing (SR) from some network diagrams, grey haired optical veterans would immediately recognize an SDH network… or at least a Carrier Ethernet one.
Admittedly, this is a quite polemical text, since without packet solutions, telecommunication would still be in the middle ages. As always, the combination of several technologies leads to more economic results. And as always, religious believes are not overly helpful. Router offload to OTN can now be controlled well enough to create value without operations difficulties, and ODU Flex (yes, TDM people can also learn) is a great way of carrying packets which simply don’t need to be routed. For certain applications, and in certain parts of a network, the best thing a packet network can do, is to behave like a TDM network.
With more than 25 years of experience in working in the telecoms market, Andreas is ECI's Head of Product and Corporate Marketing. He has made a career of advocating, explaining and selling the values of telecommunication solutions.