How Long Will I Be down?
The other important number that goes into calculating the cost of downtime is of course how long your servers are down. The average downtime for most companies is 87 hours per year, which represents 99% uptime—a number most find sounds perfectly acceptable—until you consider that based on the average per-minute downtime costs, those 87 hours can end up costing a company hundreds of thousands of dollars a year. Even if you think your services are up 99.9% of the time, that is still 8.7 hours a year, and 99.999%, which is considered best-in-class, still represents over $3700 in downtime risk. The sooner you know something is wrong, the sooner you can stop losing money.
Causes Of Downtime
While it seems like every week there is a new major attack on computer networks, the reality is that most network downtime is still caused by mundane issues.
- 40%: Hardware failure
- 25%: Service provider outages
- 25%: Human error
- 10%: System attacks
While it would be nice to be able to focus on one area that most often causes downtime, the reality is that websites are dependent on a plethora of systems and interactions. A failure in any one of those—be it hardware, software, or human error—can bring a website offline.
Does Monitoring Help?
One additional objection we hear when discussing monitoring is "Well, I can't do anything about it anyway." We are seeing more and more customers outsourcing their website hosting to large service providers. We even host the main Vigil website on a third-party service. If your website goes down, it often feels like there is little you can do other than just wait it out.
However, given the importance a website has in any business today, it is important to know if your service provider and internal infrastucture are doing their jobs right. Even large providers can run into issues, and if those issues keep cropping up it is imporant to know. Without proper monitoring in place, you may never even realize how much downtime you are suffering, and how much money you are losing.