Managed service providers (MSPs) are highly dependent on networks that time and again prove to be frustratingly fragile. In fact, a recent survey of 315 network professionals conducted by Dimensional Research on behalf of Veriflow, a provider of tools for modeling networks, finds that 74 percent of respondents admit that network outages significantly impact their business at least several times a year. A full 97 percent also concede that human errors generally play a significant role in causing those outages.
Given the complexity of networking environments, it’s often a wonder the networks work at all. Despite the rise of software-defined networking (SDNs), most networks today are still configured by hand, and each change to the network creates yet another opportunity for outage.
Benefits of machine learning
The good news is that a significant amount of research and development effort is now being poured into machine learning algorithms that promise to eliminate many of the opportunities that now exist for introducing errors. At the recent Cisco Partner Summit 2016 conference, Cisco demonstrated how it will soon use machine learning algorithms to keep track of all the connections between networking equipment during an upgrade process. Rather than an MSP needing to manually reconfigure the network after a software upgrade, the machine learning algorithms will automatically re-establish all the connections on the network as they were before the upgrade was installed.
That’s only the beginning. Intel plans to infuse machine learning capabilities across its entire line of processors, which would be used to automate the management of everything from smartphones to entire data centers.
Naturally, it will take several more generations of processor upgrades for all these machine learning capabilities to become pervasive. But the IT rank and file already sees the writing on the wall. A recent survey conducted by the recruiting firm Harvey Nash finds that almost half of the respondents (45 percent) expect their current jobs to be automated out of existence within the next 10 years.
Impact on managed service providers
What's less clear, of course, is the impact all this automation will have on the need for managed services. The cost of delivering lower level IT management services is going to drop to zero as machine learning algorithms become smarter, so MSPs will have to find ways to add value that goes way beyond simply keeping IT resources available.
In fact, it’s hard to imagine how MSPs will be able to add value without moving more aggressively into the realms of application development and business consulting. Beyond merely trying to keep the proverbial IT lights on, organizations will soon be much more interested in what they can do with IT to enhance their customer experience. MSPs that can provide that kind of insight will pick up IT maintenance contracts that will be embedded into the development of the application on the assumption that all the infrastructure on which that application depends is highly automated.
MSPs obviously still have time to make the appropriate adjustments to their business models. But it should be apparent to everyone by now that the entire IT industry is moving toward automating the management of both applications and IT infrastructure to the nth degree. The challenge facing MSPs now is figuring out how to evolve as those advances become widely embedded across IT environments over the next four to five years.