When we make major infrastructure purchases, we like to know that the choices we’ve made are smart ones — good for the long run. But the long run in my headline refers to another issue — the capability of a device to run for weeks or months without seizing up or experiencing degraded performance or loss of features. It is an issue that seems finally to be getting the recognition it deserves.
This is nothing new. “When in doubt, reboot” has been an almost daily fact of life for users of desktop systems from Microsoft Windows 3.1 through Win 2000. Memory leaks were often the culprit. Subtle errors in programming failed to release memory until all system memory was exhausted and the system could no longer run — until a reboot cleared memory. Recent versions of Microsoft’s desktop and server systems seemed to have overcome the need for constant reboots.
Today, we almost expect these types of problems from low-end network appliances, such as our home broadband routers. The first thing most of us do when we experience slowdowns is to pull the plug and restart, which often fixes the problem.
But when network infrastructure elements supporting business users degrade to a point where they slow down, lose functionality or hang and must be rebooted, it is definitely a bad thing.
In response to my recent column about Dell LAN switches, several readers wrote to say this was the same issue they had. A network manager from a university in North Dakota wrote in part:
“I wanted to like their switches. The price was right, seemed well built, interface was good enough. We bought some of their switches, seemed to work (okay) at first, but then the lockups would happen. It wasn’t consistent, but did seem to be consistent with increased traffic. Too much traffic for a long period of time, and they would stop passing it. We tried them on our campus side, then the dorm side, didn’t matter. We still have them in the corner, collecting dust, we take them out when we need to have a small, private network; for that they work great. It could be that we are spoiled by older Foundry FastIrons. We measure those uptimes in years, normally only being reset when we have to replace a UPS.”
Vendors listen to what customers want, and it is clear that long-running devices add to a company’s value proposition. In the past year, we’ve seen greater focus from our vendor customers on reliability and robustness. Still, there is nothing stopping users from demanding more details from vendors about how their devices are built and what kind of stress and regression testing they are put through before release.
Vendors that invest heavily in gear and personnel focused on improving quality will be happy to tell you all about it. Those that just slap their label on a box made “somewhere” by “someone” will try to change the subject.