Communication service providers' adoption of cloud-based solutions is not only blurring the line between traditional IT and CSPs, it's also fuzzing the definition of a bedrock phrase within the telco industry: carrier grade equipment.
"Carrier-grade has typically been associated with the iconic mathematical 'five 9s' reliability measure or 5 minutes outage maximum per year of a telecom service. More practically, we simply think of this as something that never fails. And in hardware and software designs, as an implementation that is fully redundant, where no single failure can bring down a system," says Michael Murphy, chief technology officer for North America at Nokia via email. "There is some cost to doing this."
Indeed, carrier class certification is "a rigorous process and emerged out of a unit of the old AT&T Bell system. Spun from Lucent, this Telcordia unit now resides inside Ericsson," says Jim Jungjohann, a 20-year telecom veteran and analyst who covers communications equipment technology at
Kingdom Ridge Capital. "Common among investors is the Osmine program that ensures that all of the network equipment used by the few remaining RBOCs (regional Bell operating companies) can be managed by the same software. In addition, NEBS [Network Equipment-Building System] compliance ensures the network equipment survives earthquakes and power outages."
Whereas everyone in telecom once understood the meaning of "enterprise class" or "enterprise grade," the translation is slipping, with some developers that target telecom mimicking the efforts of those vendors that once described their applications as "enterprise-ready." Add this increasingly vague descriptor to the stew of more than three dozen standards and, forget about not being on the same page; executives may not be in the same book shop.
Making the (Carrier) Grade
Carrier-grade has typically been associated with five 9s, but cloud is changing the equation, says Michael Murphy, chief technology officer for North America, Nokia.
"Today's looser definition of carrier-class equipment means more about total cost of ownership, integrating reliability with ease of provisioning, network management, and software. For instance Linux can now be classified as carrier grade," says Jungjohann. "Cloud-based services and virtualization have also added an increased level of complexity beyond just redundancy. In addition, the higher speed networks of today -- 100/200G fiber networks running between datacenters using 25G lasers and very complex coherent and QAM [quadrature amplitude modulation] signals -- have again raised the bar for obtaining carrier class type metrics."
Rewriting the book
In addition to seeking cloud-based components that still meet some or all traditional carrier-class criteria, CSPs tackle the reasons for this designation -- around-the-clock reliability -- via multiple means.
With cloud, CSPs no longer must treat -- and deliver -- all services equally, allowing them to pass along savings to customers, says Gheorghe Susman, network operations manager at SRVR LLC, a midsize global service provider for wholesale and retail consumer brands such as QuickCall.com.
"We have to balance the cost versus return of choosing only carrier-grade partners. Usually we try to partner premium (or carrier grade) as often as it makes sense based on the calling patterns we see from customers and optimize for least cost routing wherever possible, but without sacrificing quality," says Susman. "Having multiple providers means we can more easily create liquidity in our network for cost, while still maintaining acceptable levels of call quality and connectivity."
Adds Nokia's Murphy: "There is also discussion over the years regarding whether all services need to be carrier-grade. Is it critical that the Pokémon Go app never fail, ever? Probably not. But voice calls and 911 are expected to, more or less, never fail. Choices of reliability for different services is an ongoing consideration. However, it is also true that over time, the expectation of reliability increases. In years past, it may have been acceptable to have hours or even days of Internet outages. Those days are long gone. So too it might be that Pokémon Go needs to be fully redundant in the future."
As we all know first-hand, enterprise systems fail at times. Popular online game Pokémon Go is often offline somewhere in the world, according to the Pokémon Go Down Detector. Probably a lot more importantly though, Southwest blamed a router and its backup for an outage that will cost the airline between $54 million and $82 million, CEO Gary Kelly said this month. Google, Amazon and Salesforce are among the many cloud-based networks that failed this year, widely impacting business operations.
Prioritizing Content: From Vital to Pokémon Go
Virtualized systems allow CSPs to choose which services should be always on and which, perhaps like Pokémon Go, can afford down time in their design.
To deliver more reliability CSPs incorporate additional redundancy and hardens COTS, says Murphy. Newer virtualization tools help, he notes.
"The concept of hardening under this new IT/cloud environment is different than what it has historically been. In telco, we are used to having five 9s guaranteed on a per-box or hardware level. With the cloud, we should be thinking out of the box and looking into ways of guaranteeing telco grade availability on a system level," says Murphy. "This means we can continue to use servers that are IT grade -- less than five 9s -- but on a software level we guarantee the cloud architecture as a whole will provide the carrier grade availability or even more. This is done by a combination of cloud-native applications and highly automated management and orchestration solutions, enabling a fast response to any problems that arise."
Going beyond tech
Service level agreements are important to delivering on a service provider's promise, whether it's 99.999% uptime or less, says Justin Davis, director of enterprise sales California at CenturyLink.
"When selecting services, the number one item that should be evaluated is the SLA. This is ultimately what separates quality, carrier grade service providers from the rest of the pack. Plenty of marketing can be applied to product features and price, but the SLA is where the true commitment is made clear," he says in an interview. "If you look at a commoditized product, such as fiber, there are only so many factors that can impact price. If a service provider is showing pricing out of line with the market take a look at the SLA. Likely you'll find it outlines an uptime commitment much lower than the top carriers or what business requires."
Outside an SLA, customers are more willing to deal with technical issues if they're getting a novel or innovative service, says Christopher Cullan, director of product marketing for Business Services Solutions at InfoVista.
"[There is a] willingness to forgo some assurances in order to achieve a specific benefit. If I can make a transaction in my car that may fail, causing me to repeat it, versus waiting 20 minutes to get to my office, then perhaps I'm willing to sacrifice that reliability. However, such tolerance is short-lived," he tells New IP Agency. "Once a service or product becomes mainstream and heavily adopted the amount of forgiveness drops drastically. The result is that the market is perhaps more willing to experiment but once a service is adopted, it's counted on, just as it once was a few decades ago."
— Alison Diana, Ambassador, The New IP Agency. Follow her on Twitter @alisoncdiana or @The_New_IP