Of all the times for Costco’s website to go down, Thanksgiving might have been the worst.

But it happened, and it cost the wholesaler a lot of money—about $11 million from a 16-hour outage, according to Business Insider.

Now just imagine that happening to your association ahead of a big day—say, Giving Tuesday this week or during a major conference. Scary stuff, huh?

To avoid something like that happening, here’s where to look when trying to prevent a big hiccup at a big time (or even a small one):

Get to planning. A recent Uptime Institute (UI) whitepaper explains that the planning processes to prevent downtime share a lot of common ground with the things that you might have to do to bring a site back online, presumably with a whole lot less pressure. “Many of the steps necessary to recover from an outage are the same as those needed to prevent failures: have a deep knowledge of the facility; identify single points of failure (e.g., single-corded servers); identify critical loads; develop and refresh procedures; and conduct scenario tests, then update the procedures,” writes UI’s Kevin Heslin [registration].

Set up a staging site. Often, the riskiest time for downtime is when you’re making big changes. With that in mind, don’t put up new code on a live site. Test it first in an environment that is similar to your live one, so you can catch the problems before things go live—and get fragile. Check out the resources at SitePoint for suggestions on how to set one up. A vendor resource can prove helpful here as well.

Use a security tool to cache assets and fight off online attackers. Platforms like Cloudflare and Fastly are widely used to help distribute content and protect against risks as they arise, such as denial-of-service attacks. They have a lot of side benefits, too, such as the ability to cache heavily used content to help pages load faster. The result is that the technology can prove a huge perk even during the times of year when you don’t have massive amounts of traffic coming in.

Build for redundancies. If you know you’re getting a lot of traffic to your site, plan for it by using an additional server to manage the load. Speaking to CIO Dive, Yaniv Valik, vice president of product, cyber, and IT resilience at Continuity Software, noted that “load balancing,” a technique used to take pressure off the server, is but one part of a server management strategy. Mission-critical safeguards include everything “from redundant power supplies and network connections at the server level, through redundant core networking and storage infrastructure, all the way to the use of clustering, load-balancing, [and] elastic computing,” Valik explained to the outlet.

(PeopleImages/E+/Getty Images Plus)