News Stay informed about the latest enterprise technology news and product updates.

Refresh cycles lag as enterprises retain more legacy network gear

Half of networking devices are aging or obsolete, a new report says, as fallout from the recession continues to disrupt legacy network refresh cycles.

Ongoing budget constraints resulting from the Great Recession have disrupted the normal refresh cycle in enterprise networking gear -- with half of network switches, routers and wireless devices aging or obsolete. That's the highest rate seen in six years, according to a recent report from Dimension Data, a cloud and managed services provider.

Yet that statistic isn't exactly cause for alarm. The report also indicated that legacy network devices experienced hardware failures and software problems at about the same rate as newer devices did -- and obsolete gear even failed less often.

"If I were an end user of technology, I would say this is good. This is validation that I can 'sweat' the assets longer and maybe consider taking on risk," said Andrew Lerner, a research director at Gartner Inc. "I don't necessarily have to [immediately] upgrade when the equipment goes end of life."

The findings come from Dimension Data's Network Barometer Report 2014. It's based on information collected from enterprise customers that use the company's "Technology Lifecycle Management Assessment" service, which includes an inventory of all network devices. The results -- an aggregate of 288 assessments done last year -- cover 74,000 network devices across the world.

When Dimension Data first produced the report in 2008, 43% of the networking gear used by its enterprise clients was aging or obsolete. That dipped the following year to 35%, but it has consistently climbed since then to its current and highest rate of 51%, up from 48% in 2013. There are several drivers behind this trend, but the largest by far is the economy, according to Raoul Tecala, Dimension Data's director of business development for networking.

"Prior to the global financial crisis, when a big vendor like Cisco or HP would announce that a device was past end of sale, most clients would enter into a standard operating procedure of, 'Over the next one or two years, let's look at refreshing that technology,'" Tecala said. "We feel that cycle -- that mindset -- has been irrevocably changed."

Software-defined networking (SDN) was also cited as a reason, albeit a secondary one, why more enterprises are delaying network upgrades.

"SDN is not necessarily why people are holding off on their refresh cycles, but I do see it as a contributing factor," Tecala said. "Why would I want to go in on a big, multimillion dollar refresh if in 12, 18, 24 or 36 months -- whatever that happens to be -- SDN totally changes the game? So we do think there's a wait-and-see effect that probably wasn't as big in the past year but will have a growing influence in the next couple years."

Upgrades not driven by failures

Dimension Data defined "aging" devices as those that are past end of sale, which the company says typically occurs when the equipment is three to five years old; it categorizes "obsolete" devices as those past end of support, which Dimension Data puts at roughly five-plus years old. In addition to its findings about the amount of legacy network hardware out there, Dimension Data found that obsolete devices had the lowest rate of hardware failures and software problems -- 2.95%, compared to 3.78% in aging devices and 4.14% in newer devices. And the newest devices also had the longest mean time to repair, 4.2 hours, as opposed to 2.7 hours for aging devices and 3.3 hours for obsolete ones.

If it's not impacting users or impacting production, there's a sense of, 'Why spend more money on it?'
Jess Probascosenior network engineer, American Public Media

Gartner's Lerner disagreed that three to five years qualified as aging, noting that the useful life of most networking equipment is typically closer to seven to 10 years, with gear deployed at data centers spanning five to eight years. Enterprises often mistakenly treat a vendor's end-of-life notice as "gospel," he added, when they should also consider manufacturers' reported mean time between failures.

Lerner agreed, however, that the recession's fallout has persuaded enterprises to hold onto their legacy networking equipment for longer than before.

"You have the macroeconomic issues, so people stop spending … until the wheels fall off," Lerner said. "But actually what that did, in addition to just [enabling IT departments] to save money ... they saw, 'Hey, we didn't upgrade this thing when it went end of life or end of sale, and we didn't go off the cliff.' It gave them a little bit of an appetite to take on that risk in extending the refresh cycle."

Jess Probasco, senior network engineer at American Public Media, the second-largest public radio broadcaster in the U.S., estimates that until two years ago, half to three-quarters of his network devices were aging or obsolete. In the latest upgrade, he replaced one of his Cisco Catalyst 6500 core chassis switches with a Nexus 7000 in addition to swapping out about 20 of his Catalyst 3560 access layer switches for Catalyst 3850s. The older devices were about eight years old and hadn't malfunctioned, but they hit their end-of-support dates.

"I would say that [amount of time] is probably on the longer side. Typically, with hardware, it's more like three to five years, but more people like to push it [to or past] five years now," Probasco said. "[Our legacy switches] worked great. They did the job we needed them to get done, but they were going end of support and were in a production environment where we depend on them."

It's rare to find an IT department that has the budget to upgrade to the latest networking product simply because it's shiny and new. If legacy equipment isn't broken, as was the case at American Public Media, then there's often no impetus to upgrade unless the devices are mission critical and no longer receive vendor support.

"For the most part, switching hasn't changed in the last 10 to 15 years, especially in Layer 2 networks," Probasco said. "So if it's not impacting users or impacting production, there's a sense of, 'Why spend more money on it?'"

More often, Probasco finds upgrades driven by the need for certain features and functions in newer products, such as access switches with stronger Power over Ethernet capabilities to better support future 802.11ac Wave 2 products. Gartner's Lerner also noted that innovation occurs faster in some areas of networking than others, a key consideration when determining when and where to do a refresh.

"There's a ton more innovation in the wireless space than in the campus switching space, for example," Lerner said. "If you're sitting in the office and plugged in, one gig is more than adequate and one gig switches have been around for a number of years … whereas if your access point doesn't support the newer protocols, you're going to [have] laptops, tablets and smartphones that don't [perform as well]."

Dig Deeper on Network Infrastructure

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

This isn't terribly surprising. Failures happen when things change, not necessarily just because of age. When people find a configuration or architecture that works, they don't touch it. Interestingly, this is why the most stable time of year is Christmas - when employees go home and don't touch anything. I do think that some of the delays though are tied at least in part to SDN, just not in the way that people think. Decisions get delayed when there are more options. This is why fast food restaurants typically go with more limited menus. People move faster when they have fewer choices. SDN and brought a lot of new players into the space, which actually increases choice. This will necessarily lengthen the evaluation cycle, even if people are only considering one or two additional players beyond their incumbent.
That's funny about Christmas -- never heard that before, but it makes sense. One of the other sets of stats that didn't make it into the story (and relates to what you're saying) was that human error was the #1 cause of "incidents," accounting for 32% of issues with networking gear. I think it's something most people feel is fairly obvious based on personal experience or anecdotes, but it was helpful to see an actual stat on it. Telco/WAN failures were #2 (22%), followed by issues with the physical environment (e.g., power or cooling loss; 15%) and then hardware failure in fourth place at 14%.
Ii wonder if another reason for more stability is older devices is the maturity of the codebase. Everyone complained about buggy code when Cisco first rolled Nexus back in 08/09.