On the surface, it’s fairly innocuous advice when trouble sets in on some technical gadget or system: “just reboot it”. And that simple act does frequently get things back in order. But there’s more to the story, and the stakes get higher depending on what specific component is being cycled in the name of corrective action.
For whatever reason, it’s been a busy few weeks for the reboot thing on my end. At home, I’ve had to reboot my ancient cable modem to get it back operational on the provider’s network. I have an older Jeep with a fussy automatic transmission shift control that sometimes requires the vehicle to be pulled over and restarted. The list goes on, from the front yard Christmas light strobey thing with a microprocessor in it to a family member’s smartphone. When trouble hits, reboot. Nothing new or exciting here, and as a consumer society, we’re all fairly tolerant that rebooting is just part of life as “things” get stupid from time to time.
But when it comes to expensive, supposedly high-end networking components, should we have the same tolerance for the need to reboot as a “fix”? At home, the impact of the reboot is usually measured in terms of a few people and a handful of non-critical devices. At work, depending on the environment, it may be a completely different ballgame where being coerced into rebooting an important network component can make you wonder if you chose the right vendor while the senior network managers wonder if you know what you’re doing. And in the background, hundreds of clients might be stranded. Some examples are in order- again, from my last few weeks.
- Some features in the market-leading WLAN system require a reboot of an access point to get a config change to “take”. Kind of a pain, but at least you decide when to make the change and reboot the AP.
- Same market-leading WLAN system… on one of the “stable” code versions you may run into an issue on their flagship APs where traffic in a cell stops for all users. The fix? Reboot the AP. There’s no way to really know the condition is coming- just wait for the cries of pain.
- More of the same: specific user traffic on the flagship APs may stop if the relationship between radios on the same or different APs goes wonky in just the right way. The cure? Boot the client off and make the association session start over (after the user complains).
- Then there’s the market leading cloud vendor’s switch embarrassment. On a specific model of switch, when a code upgrade gets done, the switch doesn’t restart. It goes dead and needs to be manually power cycled. Given that this product set is marketed as a great fit for remote locations where you are not likely to have IT staff present, the condition really, really irks me. As in pisses me off greatly. Like, steam coming out of my ears pissed off.
Maybe it’s a generational thing, and I’m just getting older. I struggle to come up with any other conclusions on why so much performance suck from developers is tolerated by vendor exec teams. Maybe the enterprise vendor bigs are too busy lining up the next Hollywood type keynote speaker and the after-party musical headliner for the annual conference to care about things like quality and customer pain. Maybe to young developers the “just reboot!” mantra is an acceptable answer when hundreds of my clients are impacted by system-level problems, as these same developers grew up with the “reboot” paradigm in their networked homes for far longer than us older types. Maybe to them, expecting customers to reboot to deal with enterprise system deficiencies is perfectly normal and they were never briefed on the differences between the consumer space and enterprise systems when it comes to people expecting to get what they paid for.
Or maybe it’s a brilliant revenue-generating strategy: build in problems, and then charge licensing and support costs to people like me for the privilege of getting access to answers like “reboot it!” Everyone wins- except the actual customers and their end users.
To close this rant: I’m willing to occasionally reboot my consumer-grade gadgetry, but that allowance generally does not extend to work where real dollars get spent on beefy equipment . Sadly, too much enterprise-grade networking gear is starting to feel like it belongs on the shelves of Wal-Mart based on its code quality.
We deserve better, and rebooting isn’t a fix- but changing the culture that leads to the over-used need to reboot critical network building blocks would be.