Is Five Nines a Winning Hand for UCaaS?

Phil Edholm of PKE Consulting has a really great post up on No Jitter, in which he examines downtime in UC as a service (UCaaS). He provides a detailed look at a few downtime scenarios, which is an interesting exercise in itself: As UCaaS services proliferate, enterprises must understand the most common and likely downtime scenarios, and the probable effect these could have on their business operations. Believe it or not, I haven't seen this thorough an examination of downtime in UCaaS anywhere else.

I say "believe it or not" because downtime used to be an obsession in the telephony world, and it's certainly still a vital metric in wide area network (WAN) service level agreements (SLAs). For telephony services, the magical metric was, of course, "five nines" -- 99.999% uptime. When voice was first being sent over IP networks, we ran countless posts on No Jitter and innumerable sessions at Enterprise Connect discussing how close the new VoIP technology was to attaining five nines. That it needed to reach this pinnacle was never really questioned.

For those who don't come out of the glory days of the public switched telephone network (PSTN), or who haven't happened to run into this metric elsewhere, a network with annual uptime of 99.999% is one that only suffers 5.26 minutes of downtime a year. In other words, a single six-minute network outage eliminates the possibility of five-nines compliance in the year when it happens. That's a stringent threshold, and one that the Bell System telephone network monopoly pursued with a sense of mission throughout the 20th century. It was matched in the PBX industry, as Phil points out in his post:

    If a premises PBX vendor touted availability of three to four nines (only one to nine hours of downtime per year), that vendor would have been challenged to sell many systems against competitors claiming five nines. On the other hand, as Curtis [Peterson of RingCentral] noted, on legacy systems we can't track issues as transparently as we can with cloud-delivered services. Regardless, customers considering a cloud migration need to understand availability of their chosen provider and the impact on their organization.

As Phil points out in his article, the number of elements -- peering partners, hosted datacenter providers, and others that may be involved in actually enabling a UCaaS service -- makes achieving five nines in UCaaS a real challenge. Five nines was feasible in a time when there were fewer elements, with fewer entities controlling them. The traffic originated on your PBX, transited the PSTN, and terminated on a similar PBX at another enterprise, or at an individual phone hanging off an equally-robust public-network switch. Or the call traversed private lines between PBXs at your own locations. Each of those (relatively few) elements could offer five nines, and if they all lived up to the promise, the service was rock solid.

The ultimate question here centers around Phil's final point in that quote above: "the impact on [the] organization." In that world where the PBX was the only game in town for voice communications, five nines was a must, because there weren't that many other ways to reach people in real time. But now, landline calls may not be as mission-critical: Say you've got several dozen remote offices hanging off a UCaaS service, and a localized outage -- whatever the source -- knocks out landline voice to one of those offices for an hour -- but this may be an office that only gets a handful of "traditional" telephone calls each day anyway. Because you're a business, you really can't deny your potential customers any connectivity option -- even the landline telephone. But the impact of an outage may not be nearly as great as, say, a public cloud outage that takes down the business's communications platform as a service (CPaaS) service that's providing text messaging services. And the remote office that loses its landline telephony may be able to go on functioning via individuals' cell phones without missing a beat.

So the question is, "What are we willing to give up to get five nines?" No matter how good any individual service provider is at any point in the chain, the more elements a service has, the more vulnerable that service becomes.

Whatever the priorities are in any given situation, what they must always be is yours. If UCaaS is going to replace legacy PBXs, and if voice calling still matters, you may decide you're not comfortable giving up too many nines. But you may also want to make sure you've got strong SLAs on all of the services you're buying from providers across the range of voice, video, and data.

Follow Eric Krapf and No Jitter on Twitter and Google+!