7 Strategies to Contain Network Costs (Layer 6 Will Amaze You)

First, thanks for indulging the clickbait title joke. Serialization is the unsung hero of harmonizing network and application relations and deserves the occasional, snarky callout. Moreover, identifying how the unique mix of network clients in your environment consume your carefully manicured infrastructure is critical for managing network cost. Because today, a rapidly expanding, diverse pile of new technologies all assume the network is a magic grid, no tuning required. Instead, IT pros are the overloaded “magic” grid and finding time to identify potential network cost savings can be difficult.

In this article we’ll review seven proven strategies to help contain network costs and make network engineers even more valuable to the business along the way.

  1. Accept That Hybrid Is Forever and It’s a Legit Pain in the SaaS

    Let’s get this one out of the way. Hybrid is what happens when unlike things are mashed together. It’s not a strategy, there’s no invoice SKU for it, you didn’t buy it from any vendor. Instead, you accepted it as a new cost and operations responsibility to add cloud, HCI, SaaS, edge, and everything else anti-premises. For network engineers, the greatest challenge is it breaks twenty years of progress integrating network hardware, protocols, and tools. SNMP is a bit long in the tooth, but it’s wonderfully ubiquitous. Point a mature network management system at network gear—even in multi-vendor hardware environments—and teams can quickly discover and begin basic service monitoring. Well, until recently anyway.

    Now, with hybrid environments, teams not only wrestle with assuring new investments in technology work, but the basic process of monitoring is also becoming more complex and expensive. Hardware vendors are bundling proprietary “free” tools into invoices to fill ops observability gaps, but by transferring cost of new tools for each technology to buyers. The whole point of IT pros prior streamlining efforts has been to eliminate swivel-chair integration, and hybrid is reversing this trend.

    Break out the books on your current NMS and look for new hybrid monitoring capabilities in recent upgrades which might have slipped under the radar. The goal is to get all your networks back into a single view regardless of where they are or even exist in the physical realm. Fewer tools means more savings from team efficiency, and fewer parts to manage over the long run.

  2. “Bandwidth” Is Not a Physical Property, It’s Taxonomy

    The amount of bandwidth you’re billed for every month doesn’t matter to end users, only how you deliver it. For example, have you ever doubled a link’s provisioned capacity only to see it immediately peg the gauge right back at 95% utilization? The sinking feeling many of us felt watching this happen stems from experience that new traffic cannot all be business traffic. Only the network cost is guaranteed to increase, not service quality. Technology like SD-WAN can help, in some cases cutting network bandwidth costs 60 to 90% over traditional MPLS leased lines. However, to realize those savings without affecting the end-user experience, you’ll need to analyze and categorize the real-world mix of application and user behavior of your network.

    Look to flow analysis tools to turn firehoses of NetFlow and sFlow logs into useful data to support bandwidth spending decisions. Better, use traffic analysis to underpin routing configurations for SD-WAN, QoS, and other traffic shaping and optimization protocols. Not every byte has the same value to the business. More of them can flow over less premium links than you might think. Also, consider adding a bit of multipath monitoring to your network dashboards.

    Even though you can’t fix issues in your ISP or cloud provider’s networks, tickets will still be opened in your service desk when externally hosted infrastructure isn’t performing well. Multipath visibility can help IT pros be more proactive in resolving external provider issues, reducing network support costs.

  3. Shrink the Data Center

    It feels like heresy to suggest network engineers can be catalysts for cloud or SaaS adoption, but it’s true. Public cloud providers now represent both the largest enterprise IT infrastructure in the world and the largest MSPs in the world. Letting someone else manage tedious, routine network operations tasks can help your team conserve resources and costs. Your first thought may be, “But the data center is 75% of the jobs in networking outside wireless and telephony. Why outsource my own role?” TL;DR you won’t be.

    First, cloud still has networks because without networks applications would just sit there, furiously burning OpEx by the hour. The advantage for net admins is a cousin of automation. Cloud networks are more efficiently managed using APIs and software, not complex command-line configuration. However, to be successful your performance visualization tools can’t go dark as the infrastructure becomes opaquer to monitoring. Investments in cloud monitoring remove both migration risk and ongoing cloud network operations cost, just as it does in the on-prem data center. The only thing you may miss are a few thousands steps per day roaming hot and cold aisles with a crash-cart.

  4. Consider New Vendor Options for Campus Networks

    One aspect of network that won’t be disappearing any time soon is delivering services on-campus. Perhaps in your environment, you’re already seeing a shift where WAN and LAN distribution networks increasingly represent a greater percentage of hardware than data centers. That’s in part because of dramatic improvements in infrastructure service concentration for core and aisles, cloud migration, and wireless expansion inside the office.

    New players—and reenergized, storied brands—are increasingly focused on cloud-managed, low-complexity gear. Re-evaluate your vendor list regularly for opportunities to reduce cost, improve service for existing infrastructure, or save team time to better invest elsewhere on the network.

  5. Consider Stretching Hardware Refresh Cycles

    Maintaining conservative hardware refresh cycles goes a long way toward helping network professionals sleep better, especially on the weekends. Even when “wasting” a bit of serviceable life, few enterprise infrastructure elements can buy their way to proactivity like network gear. It’s good for IT pro job security to ensure critical routers and switches don’t catastrophically fail despite lovingly applied layers of duct tape. Also, it’s important to populate your network with hardware capable of addressing the challenges of modern workloads and user expectations. Sooner or later those dumb switches hiding in dusty closets have got to go.

    The challenge is a good chunk of deployed gear today can be useful longer than ever before. There’s significantly more flexibility, modularity, failover redundancy, and service maturity on the wire than even five years ago. So, how do admins separate useful cold-rolled steel from aged network chaff? Use your network management system in collaboration with inventory management and ideally your service desk to analyze common longevity challenges. Are there vendor or model types with higher failure rates? Which elements more gracefully mature with user demand? Where are you investing more time doing unplanned break-fix? Can you repurpose existing gear to serve new roles? Your network can help identify a more a cost-effective mix of hardware refresh strategies. One cycle does not fit all.

    One note. Extending the life of hardware can increase operations risk without a bit of additional monitoring. Look to your network configuration management tools to help identify expected end of support, end of life, or security vulnerability details. This information simplifies decision making and can help convince management you’ve got risk under control.

  6. Measure Shiny Twice, Cut Once

    Shiny is glorious. Every network engineer at some point has stood facing a rack, with electric tingles from the fabulousness of newly installed dream hardware. I know I have. However, for cost containment, it’s critical to be wary of seemingly miraculous new tech, because it often drags rip-and-replace—the most dreaded cost monster of all. Yes, truly transformational tech is worth its weight in plated gold to the business. Confirming this, however, is up to the network team.

    Shiny is fun to work with, and as long as the new hardware makes a difference in team efficiency, lowers potential incident cost (risk) with improved security, or drives growth via user sat, then by all means acquire some. But if elements of your current network are good enough and you can prove it, hang on to what you can. And if you need reassurance, just ask your CFO. She’ll be happy to reassure you that less-cool is fine with the business.

  7. while (true) { Automate(); }

    The cost containment advantages of automation cannot be over-stated. It’s impossible to over-state the value of automation to control cost. Perhaps more than any other single factor, network cost containment benefits from automation. break;

    The wheel is good. Fire is good. We have air conditioning today because as a civilization we strive to remove toil through technology to make time for higher level work. Your network is no different. How many of your daily tasks are routine? For most network engineers, it’s more than half of the time spent on a typical day. Fortunately, network engineers don’t need to be programmers, and many would prefer not to be anyway. However, as packed Cisco DevNet sessions demonstrate, even senior network engineers are eager to learn how to code.

    IT pros increasingly turn to code to create more complex automation tools than is possible with basic scripting. Beyond that, network management vendors are increasingly including built-in automation capabilities, designed specifically for network operations optimization. If you haven’t investigated the automation capabilities of your NMS, do a quick search of its documentation for the word “Automate.” You may be surprised how much functionality is already available.

    Most importantly, research the benefits of automation training in other enterprises and your competitors if they’ve spoken about it. You’ll see one leadership focus repeated in the most successful IT teams: they invest in automation-specific training and research. While useful automation does not require programmers, introduction to code can have a steeper initial learning curve for admins with decades of green screen experience. They quickly get the hang of it as they do with other new technology, but the first step—the first working HelloWorld() automation tool—can take time. Investing upfront in automation and tools with automation features can pay huge cost containment benefits later.

Adding Cost Containment to Your Networking Guru Toolchain

Hopefully these strategies have your wheels turning, renewing interest in network cost containment opportunities, and imagining new ways to meet business goals for budget trimming. The exercise can be much more worthwhile than it might first seem. Not because it will please the CFO or CIO—which it most certainly will—but because it helps return local control of IT spending back to IT pros.

When the business knows from objective data that IT is continuously looking for opportunities to save money, there’s more autonomy to invest where it’s really needed, in training, in overlooked upgrades, and perhaps for a few more shiny things to play with too.

Anonymous