Before Signing on the Dotted Line for Higher Uptime

Paul Mah

Previously, I discussed whether small and mid-sized businesses really require 99.999 percent uptime, also known as the "five nines." I highlighted that many SMBs specify an arbitrary uptime based on flawed notions of perfect availability and a failure to understand the limited impact of occasional downtime.


Central to the topic is the exorbitant cost of maintaining a higher level of uptime, which goes up exponentially with each additional "9." As a result, it might not only be unnecessary, but also impractical for the limited resources of many SMBs.


On the other side of the fence, though, are companies where obtaining the highest availability is central to the viability of their business. Today, I want to look at some factors that SMBs working toward a high level of availability should examine more closely before signing on the dotted line.


Is the Provider Really Offering 99.999 Percent?


I know of hosting companies that will offer "100%" uptime with the explicit promise that they will reimburse part of the monthly hosting fees if they fall short of their guarantees, upon request. And of course, these providers will typically charge slightly higher fees than the industrial norm to better reflect the cost of maintaining their superior infrastructure or hardware, which boosts the SMB's confidence.


However, many SMBs don't realize that this is in most cases nothing more than a marketing gimmicks. For one, any "damages" is limited to the cost of the monthly subscription, and even that is sometimes pro-rated to the duration of the downtime in the fine print. In addition, these promises also tend to rely on the SMBs finding out about the service lapse in the first place, and then jumping through various administrative hoops for the (often limited) fee refund.


As you can imagine, the problem is that SMBs often don't notice when availability guarantees are breached - not without a catastrophic outage or third-party monitoring at least. (Indeed, this is one reason why I have respect for organizations that provide updates and refunds from service lapses without demand.)


Beware 'Scheduled' Maintenance Loophole


As with all technological products, there is a need for periodic maintenance for various software updates or upgrades. This can range from patching operating systems, upgrading hardware, configuration changes or infrastructure upgrades. To deal with such eventualities, most providers to include downtime due to scheduled maintenance in availability guarantees.


So let's say your hosting provider hosts your website using a Windows-based server and has to restart an average of six times in a year after installing Patch Tuesday updates. Assuming the server takes about 30 minutes to complete installing the patch, restart, and properly re-initialize, that's already three hours of downtime over the course of the year that's outside the contractual terms that you signed up for.


So if you really need 99.999 percent uptime, you might want to specifically work out an arrangement for load-balanced or redundant systems in which some systems will always be up even when maintenance is in progress.


Perception of Uptime


Unless your business conducts business solely through the Internet, the trust is that the occasional downtime will not kill your business. Consider this: Popular social-networking site Twitter was down for 52 minutes last month and 37 minutes in April. Assuming similar performance over the course of the year, this works out to be about 45 minutes of downtime a month, or barely 99.9 percent uptime. While some might argue that Twitter can hardly be considered business critical, it is also a service that is literally accessed around the clock - and its performance this year is already much better than last year's. Yet users are hardly leaving the social-networking service the last time I checked.


So is system availability important? My answer is yes, without a doubt. Unfortunately, uptime numbers have been skewed or abused for years in the name of marketing. And while it is certainly desirable to have the highest uptime possible, it would be far more prudent to first evaluate the minimum requirement of your SMB and work toward higher availability based on what you can afford.

Add Comment      Leave a comment on this blog post
Nov 10, 2010 11:34 AM Terry Terry  says:

I would argue that you should evaluate the cost of downtime in lost productivity, revenue, and reputation and purchase the solution that meets your needs. 

Just ask the vendor how they plan to deliver that level of availability and get information on the server and network hardware.  There are a limited number of vendors who can really deliver.  If it's a true fault-tolerant server, software can be hot updated, meaning they don't even have to be brought down for upgrades (planned downtime), so they only time they are down is if the power is out and the generator and UPS fails too... or an actual disaster occurs.


Post a comment





(Maximum characters: 1200). You have 1200 characters left.



Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.