A few days ago, Magento released their eagerly awaited 1.1.7 update. Not long after, the Magento Forums began filling with the usual trouble reports, the type one expects with any new release. The sad part was that a good number of these reports were from people who had installed the update on a live store and who had subsequently encountered deal-breaker problems that entailed reverting to an old version. Now, some people – myself included – might question the wisdom of installing a software update to a live e-commerce store just days before the holiday rush hits. And they’d be right. But the overarching question is whether they should have installed it at all and whether they could have avoided the trouble if they’d approached the upgrade with more caution and, certainly, better timing.
I use this Magento example to illustrate my point, but I’m really speaking in a larger sense about software upgrades in general, with the one mentioned being just the most recent example. By now the memo should have gone out: bugs happen, so don’t trust new software, no matter how competent the creator. And I’m not faulting developers: bugs are a way of life; it’s how we deal with them and, more importantly, plan for their existence, that defines how much trouble they can cause.
In recent years I’ve begun to take a very hard look at application and operating system upgrades with an eye towards answering the question, When are upgrades appropriate? My conclusion is: less often than you might think or be trained to believe, and even if you take the plunge, delaying it can often be the wisest move you can make. The Devil You Know is easier to deal with than The Devil You Don’t.
Before your next software update, ask yourself the following questions:
- Do I Need This Update? What is to be gained by installing this? If you’re not getting something important, like a critical security patch, a desperately needed feature, or major performance gains, chances are you’re wasting your time. Don’t install software updates just because they’re there.
- Can I deal with the necessary recovery time if things don’t go as planned? Always plan for the worst case: an update that breaks your site entirely, leaving you no choice but to revert to a backup. Is the time it will take to recover from an upgrade gone awry acceptable?
- Can it wait a while? My policy is always – always – to wait at least a day or two before installing an update. In recent years I’ve saved myself a lot of grief just by holding off until the first batch of early-adopters get their hands on the latest release and put it through its paces. In at least one case – with Mac OS X – I saved myself from installing an update that would potentially have caused severe disk corruption.
- Have I taken the necessary precautions? Have you backed up all your files and your databases? (this is another best practice, whether you’re performing an upgrade or not.) Do it! You need to leave yourself a trail of breadcrumbs so you can extricate yourself from an upgrade that goes badly. Without backups you’re going to have a very difficult time doing that.
- Am I doing this at the best time of day? When is it that upgrades and, if necessary, recoveries will have the least impact on your users? Chance are that it’s going to be over the weekend. Though no one likes it when I say this, more often than not the best time to perform any upgrade is going to be on a Friday evening. Why? Because site usage will be at its lowest and, should something go wrong, you have the entire weekend to fix it. Conversely, the worst time to do an upgrade is going to be Monday morning.
The unwritten rule is to allow for the occasional exception. Just be sure the exception is for a sensible reason.
Being cautious or, as I like to think of it, prudent, is never the most exciting path. But I contend that the type of excitement that results from being the first kid on the block to run the latest software is the kind of excitement you just don’t want or need. And, often, it just isn’t good for business.
