IT teams today are trying to outrun decades of technical debt when orchestrating enterprise networks. Well-intentioned policies, piecemeal approaches, and outdated infrastructure have all combined to create a network that is bloated, complex, and prone to disruption from misconfigurations and security threats.
To overcome these challenges, teams of highly skilled network administrators are relegated to the tedious and time-consuming work of simply keeping the network up and running. Every moment spent maintaining the network is another moment that those highly skilled — and highly paid — network administrators could have spent improving the network and contributing to the enterprise's bottom line. Many companies have tried to implement in-house, custom scripting to begin automating away these tedious tasks. However, these projects nearly always falter under the weight of their own complexity and leave companies disillusioned with automation altogether.
The promise of automation is more real than ever, but it won't come through the outdated approaches that got us here. To reach the potential of network automation, enterprises must step back and recognize that the solution is as much about the process as it is about the technology. By understanding the origins and costs of unsustainable network management approaches, enterprises can apply the lessons from other technology evolutions and finally initiate a network automation project that achieves the goals they're after: a network that is agile, scalable, efficient, and resilient.
An Unsustainable Approach
Managing networks becomes increasingly challenging and costly as enterprises scale. Manually addressing basic network management tasks like configuration changes, moves/adds/changes/deletes (MACD), and OS updates requires time-intensive effort from skilled network administrators. Even routine device changes can be costly, with the scope of the problem becoming clearer once this time is multiplied by the number of devices to be managed. Cisco found that 95% of network changes are performed manually, resulting in operational costs two to three times higher than the cost of the network.
Automating these processes through one-off scripting or larger DIY automation initiatives introduces even more cost and risk. Each device type, operating system, vendor, etc. requires highly skilled engineers to adapt their automation scripts and approaches. This quickly leads to inconsistencies in the code across the network, which is a recipe for expensive outages that are hard to trace and fix. Depending on the size of the enterprise, the costs of this downtime can range from thousands to millions of dollars per hour according to Gartner.
Those who have built a custom home know that DIY projects never go to plan. The problem enterprises faced was akin to trying to build a "smart" house by having a group of builders start at each corner of the house's foundation and try to build toward the middle. To make it more difficult, each team of builders speaks a different language, and some of them will leave halfway through the project because the neighbor is paying more. The best result one could hope for is a leaky roof — you can be certain it won't be a "smart" house, and you'll be lucky if the whole thing doesn't collapse in on itself.
The cost of committing specialized talent to these tedious projects makes it virtually impossible for network operations to scale efficiently. In sum, the costs, time, and risks that are caused by manual network management and in-house automation programs are unsustainable, slowing enterprises down and inhibiting their ability to respond to new business needs.
More Money, More Problems
The unruly state of enterprise networks today can be traced back to the lack of a comprehensive solution over recent decades. As networks became increasingly complex, various point solutions appeared to attack new problems as they arose. As these point solutions piled up, enterprises had to hire additional experts to implement and maintain these tools. Eventually, coordinating these disparate tools across increasingly complex networks became too much to handle. In response, network administrators turned to DIY automation programs that leveraged custom scripting to alleviate the growing burden of basic network orchestration.
Years of technical debt, finding and retaining talent, and inconsistent scripting approaches all contributed to these projects' failures. No matter how much money was thrown at the problem, nearly every network automation initiative that relied on custom scripting alone left the network more convoluted than before.
Drawing Parallels
The similarities between the emergence of intelligent network automation and the evolution of enterprise storage management technology provide a clear case study for the near future.
In the early 1990s, enterprises were stuck with a complex, manual, and piecemeal approach to storage management. Backup operations were heavily tape-based, requiring extensive manual intervention to handle tape rotation, provisioning, off-site vaulting, and restorations. Rather than using out-of-the-box solutions, many enterprises had to build their own backup systems by stitching together disjointed tools and scripts. Scalability was also a major challenge as data volumes outgrew traditional storage architectures, leading to reliability and performance issues. Finally, ensuring stable and resilient storage with features like failover and replication was complex and depended on deep in-house expertise. The hope for a useful storage management solution had slunk into the trough of disillusionment as companies devoted more and more resources only to receive lackluster results.
It wasn't until the mid-1990s that Veritas Technologies pioneered a new storage management approach that enabled policy-based automation and pre-built solutions with extensibility and cross-platform support. Veritas' focus on ensuring turnkey, scalable, and resilient solutions significantly reshaped the enterprise storage landscape and single-handedly dragged the technology category up the slope of enlightenment. In doing so, it provided a roadmap for solving the network challenges we face today. Enterprises no longer build data storage management software; they buy enterprise software. Similarly, enterprises can now buy software for network automation instead of building it themselves.
Network automation has also transitioned from point solutions to integrated platforms, from infrastructure-centric to intent-based policies, and from custom scripting to standardized no-code development. Platforms that offer these core automation capabilities out of the box will unlock easier adoption, faster ROI, and substantially lower risk. As seen through Veritas's journey, this is the promise of effective commercial platforms, enabling enterprises to leap over traditional orchestration and automation challenges and focus on more strategic efforts.
Studies by Enterprise Management Associates show that adopting effective network automation can reduce maintenance times by 79%, and Gartner estimates that it improves operational efficiency by over 30%. To move on from the piecemeal approach of the past, enterprises must acknowledge that successful automation projects are determined not just by the strength of technology but by having a codified way of deploying it.
The Tipping Point
Many enterprises want to initiate change in their networks but find themselves overburdened with the technical debt. This is unsurprising when considering many enterprises' IT environments include decades of inconsistent custom scripting, abandoned DIY automation programs, and legacy processes that stall progress.
While traditional network automation approaches left many companies disillusioned, it is clear in hindsight that those disjointed and expensive custom approaches could never achieve the scale they sought. Fortunately, the evolution of storage management solutions provides the roadmap for success. Platforms that offer core automation capabilities out of the box will get network administrators out of the trenches and enable companies to advance the network's business value through greater operational agility, efficiency, scalability, and resilience.
I'm confident the platform that will drag network automation up the slope of enlightenment already exists; we're at a tipping point. However, the proof of this tipping point won't be contained to humming server rooms; it will be found in raving customer reviews, in expanding business units, and in overperforming financial results. The masses won't be far behind once the early adopters begin demonstrating these benefits, so the time is now to decide to which camp your enterprise belongs.
About the author:
Matt Westover is Gluware's Chief Revenue Officer. He leads all aspects of Gluware's sales, business development, and strategic partnerships. Matt has over 25 years of executive experience in information technology and network infrastructure, and a proven track record of leading startups and medium-sized companies to achieve exponential growth. Most recently, he served as Sales Vice President of Eastern U.S. and Canada at SUSE (SUSE.F), a $600M+ enterprise open-source Linux and Kubernetes software company. Matt spent the bulk of the previous decade at F5 (FFIV), a $2B+ multi-cloud application services and security company committed to optimizing and securing every app and API anywhere, including on-premises, in the cloud, or at the edge. While at F5, Matt led the financial services vertical accounts teams to become one of the fastest and highest growing businesses at the company.
