The evolving CSP landscape

The role of communications service providers (CSPs) and the world in which they operate have changed a lot in the last decade. Ten years ago, CSPs were primarily operating large, proprietary networks to deliver fixed-line phone service to homes and businesses. Since then, CSPs have transformed into true service providers, delivering not only IP-based landline connections, but also Internet access, mobile network connectivity, private data lines, VPN (virtual private network) services, entertainment media such as cable TV and many others.

 

With this diversification of services, demand for connectivity in general is also growing, and many operators are answering that call by dialing up their services and increasing available bandwidth loads. Take the proliferation of smart devices, for instance. According to Cisco, we can expect 50 billion Internet-connected devices by 2020 – with a current global population of more than seven billion. This means that there will be more than seven times as many smart devices as people in the world!

 

For the most part, the rapid evolution and expansion of CSPs’ businesses and networks has taken them by surprise. Therefore, many of even the biggest players are still operating their legacy networks like they were a decade ago, supported by antiquated standards or network and application performance management solutions that don’t take into consideration the real-time nature of most of today’s networks. For instance, much of the network and capacity planning executed by operators today relies upon performance data collected and analyzed by hand – a tedious, and extremely time-consuming process. Instead, CSPs should be utilizing a more modern and sophisticated method of network management that doesn’t require human involvement; it’s simply not feasible for CSPs to take such a resource-intensive approach to network management and performance monitoring while also investigating and deploying the next generation of network services.

 

So, how can they free up enough budget and human capital to appropriately research the trends that will be worth investing in – and then innovate in those areas?

Assessing the value of automation

Automation is widely believed to be the answer to these questions. This is because it can enable CSPs to, for example, download a patch onto hundreds or even thousands of network servers at the click of a button. But, what many CSPs don’t realize is that even if this kind of automation appears to save time, creating and running the patches still consumes many of their existing resources. This isn’t a solution – just a shift in resource usage. This would be similar to spending 100 hours building a robot to predict just this week’s weather. If the weather will be the same for the next year, the robot would provide some return on investment, or ROI, in the long run; but since a new robot is needed every single week to accommodate changing temperatures and conditions, there is essentially no time saved.

 

Similarly, scripting automation for every patch, or every network data collection point, may feel like a time savings; but CSPs are more likely to “break even” in the long run. This type of automation thrives in a homogenous environment, like the world with the same weather every day. And while scripted automation was actually extremely effective for CSPs in the 1990s when all 10,000 network servers were performing the exact same functions, today’s increasingly heterogeneous networks, which incorporate everything from IP to backhaul to Wi-Fi hotspots to femtocells, are much more complex.

 

Instead, CSPs need automation that can adapt and reframe itself to match their constantly changing parameters. Many are realizing this shift and are incorporating smarter automation tools into their network management. One such approach, autonomics, can actually script itself by watching engineers’ day-to-day activity, effectively automating the automation process. That means engineers aren’t spending time on OSS updates or analyzing points of network degradation; in fact, the autonomic system can operate completely independently from human engineers to perform these tasks and more. For CSPs, that can mean a 10 to 20 percent reduction in network management costs. And, if we look at what that 10 to 20 percent increase in resources can accomplish when applied to R&D or next-generation service delivery, the impact of autonomics becomes even more significant.

Putting autonomics to work in the CSP realm

As an example of what autonomics can help CSPs achieve, let’s take a look at service assurance, which is arguably the most important business outcome for a CSP. Wireless subscribers have extremely high expectations for network availability, particularly when it comes to mobile service, and autonomics can ensure those expectations are met. Take voice and video calls: by leveraging autonomics, CSPs can better monitor the status of a line – either a private circuit or a wireless connection – and, as error rates increase, automatically assess the potential root causes of those issues. Then, based off that analysis, the autonomic system can actually correct the connectivity problem, as well as adjust relevant network settings and configurations to prevent that issue from reoccurring. This might involve, for example, rerouting mobile traffic over another circuit or adjusting the angle of cell towers to provide better coverage in a given area.

 

The entire process of diagnosis and correction is thus conducted at machine-speed, dynamically and in real-time – sometimes, in as little as a few minutes. What results is a problem avoidance scenario for CSPs, who are able to get in front of network issues before they occur, and deliver on the proactive stability expected by their customers.

 

In addition to connectivity issues, operating system upgrades – particularly across large networks – are another area where autonomics can be applied. For network service providers, regular OS upgrades are crucial to maintaining revision levels and mitigating risks from known bugs and security leaks. However, given the man effort associated with updating systems on such a large scale, those service providers often need to perform upgrades in phases, often via contract labor. Over the course of a year, the network is slowly upgraded, one segment at a time. This means that by the time the entire network is upgraded, a new OS release is already available, driving a vicious cycle of recurring labor costs year over year.

 

Autonomics can make this a much more fluid process, and can even manage network-wide upgrades without engineers’ involvement. By essentially learning to identify which version of an OS is running on a given device, the system can determine whether that device is a candidate for an upgrade. Post-diagnosis, it can automatically identify and schedule a maintenance window, request and confirm approval, and perform the end-to-end upgrade, including all change management activities, validation, documentation and ticketing. Improving the process even further, autonomic “virtual engineers” can work in parallel, rather than serially.

 

The end result is that CSPs can compress the timeline required for OS upgrades on each device, and full network upgrades can be completed in just under a month as opposed to an entire year. Plus, autonomics mitigates the need for contract labor, allowing CSPs to focus their resources on other more innovative investments.

Barriers to adoption

So, you may think, if autonomics is such a perfect solution, then why is network management still an issue for CSPs? Actually, I’ve noticed that the move to autonomics is less about applicability and, instead, more about overcoming engineers’ hesitancy to give up control. Many engineers have been running their pieces of CSPs’ networks for five or 10 or even 15 years, and believe that the complexity they deal with is so unique that no one else could possibly manage it, least of all a machine. That’s not to say they’re wrong – after all, they’ve seen attempts at automation eat up time and resources with intensive scripting and frequent updates. What may surprise many of these engineers, though, is that autonomics can easily learn the complexities of any network, from simple, fixed-line networks to more complex 4G/LTE networks, simply by mimicking what they have been doing and applying that to like situations. That doesn’t mean the engineers are out of a job, either. Given their in-depth knowledge of the network, they’ll be invaluable in innovating new services.

 

By introducing autonomics, CSPs can begin this transition from solely supporting legacy technologies to simultaneously funding new services and sectors, like bringing LTE to new markets or building up infrastructure to support machine-to-machine connectivity. With the ability to transfer capital to new and market-leading ventures like these, CSPs can step ahead of their competition and propel themselves to the forefront of the growing telecommunications market.