What is carrier-grade anyway? It’s one of those terms that I’ve heard many times in my career, but not usually with a clear and concise definition. “Five Nines” is thrown around liberally as well, and not always with the full connotation of what it implies, but the term is at least rooted in a clear definition.
The reason I ask is that many industries are fully embracing the public cloud and all the benefits it offers, but it feels like telecommunications is taking a more cautious approach. Is this because the telco industry is fundamentally unique and it’s more of a challenge to migrate to the cloud? Or is it because of an inherit bias that running networks, and by extension data centers, is something that only telcos should still be uniquely qualified to provide?
In telecommunications of course, there is a middle path, with the promise around NFV, and the ability to achieve many of the agility benefits of the cloud but still run from a private data center. NFV also solves some very specific problems with regards to data-plane network element virtualization, and these requirements are suitably unique in that they don’t apply to public cloud compute services in the same way. Yet telco infrastructure spans far more than network resources, while the back-office IT systems and BSS represent a system landscape every bit as complex as other industries. This functionality could be virtualized in NFV, but could equally move to the cloud. Even the responsible teams (Network vs IT) are often disparate and with separate decision processes.
I’ve spoken to many customers interested in the opportunity to move their BSS to the cloud. However, there are few concrete examples and past success stories to review. People are interested, but apparently not making the leap into the cloud. So why the reluctance, is it because public clouds are not regarded as “carrier-grade?” To get a better understanding of the apparent apprehension towards the public cloud, let’s examine some of the cloud myths which may be deterring broader cloud adoption.
Myth 1 – Cloud deployments are only suitable for smaller business
I don’t think cloud suitability can be put down purely to the size of a business. Often the cloud proposition is sold to developers as, “start without capex and scale seamlessly when your product hits a growth curve” (the low capex start may even be usage of some free tier). It sounds great, but this marketing aspect provides a theme that the cloud might be most suited to “app developers” or “startups.” Does it appeal to established carriers? If you look beyond some of the marketing hype, there are plenty examples of big businesses making strategic choices regarding the cloud.
The U.S. banking firm Capitol One runs exclusively on the AWS cloud. The CIO is quoted as follows: “We recognize that we want to be in the business of building great applications for our customers, not in investing to build costly and complex infrastructure.” Capitol One’s revenues in 2016 were $27.5B, not significantly behind either T-Mobile or Sprint (the 3rd and 4th largest U.S. carriers).
Myth 2 – Having data in the cloud is less secure than a private data center
Cloud concerns could be associated with a perception of increased security risk. Unfortunately, there have been some high-profile security exposures as a result of badly configured cloud resources. In one of the highest profile cases from June 2017, a researcher discovered an unsecured database maintained by an analytics company working on behalf of the Republican National Committee that contained personal details of 198M Americans — nearly the entire U.S. electorate. 25TB of confidential information could be downloaded by anyone who chose to look. There are other examples, mostly involving AWS S3 services. In each case, it has to be pointed out the systems were inherently misconfigured, it was not a weakness of the cloud per-se. With consistent processes and policy in place there are not comparable examples of malicious actors gaining access to these systems, any more than within a private data center.
Myth 3 – Services running in the cloud suffer from instability
There could also be reluctance with regard to the perception of cloud system stability. This might be linked with “cloud is for startups and apps,” but also due to a number of high profile and well publicized outages. Well publicized, because the problem with a cloud platform outage is that it can take a big section of the internet down with it! In March of 2017 an AWS employee incorrectly shutdown more billing servers than intended, starting a chain reaction affecting multiple AWS services and numerous internet companies: sites were down from Trello to Quora. For a period of time even Nest thermostats couldn’t be controlled by their app. Though the high-profile nature of these outages create headlines because of the number of services impacted, I would counter that the cause and frequency is really no different to similar situations within carrier networks.
It is clear that many premier telcos around the world offer great technology and are in many cases experts in cloud infrastructures. In the follow-up blog to this two part series, I will discuss why I believe “cloud-grade” is on a par with “telco-grade” and why some telcos may look to leverage global cloud service providers to launch critical BSS applications.