Executive Insight: The Truth About Uptime

This piece was written by J.D. Oder, Shift4’s CTO/Sr. Vice President of R&D. J.D. is one of the original architects behind DOLLARS ON THE NET and continues to lead security and development efforts at Shift4.

There’s a lot of buzz in the payments industry about uptime and how many “nines” you should be looking for in a potential service provider. To make sure we’re all on the same page, let me explain what I mean by uptime and what exactly these nines signify.

Uptime is simply the percentage of the time a given system was up and running over the last year. For many years, the ideal that we all strove for (but few actually reached) was the impressive 99.99% uptime. We called this “four nines” (99.99) and it was an incredibly difficult standard to reach – four nines meant your system could be down no more than 52 minutes per year, or about one minute per week.

Uptime should not be confused with availability, which is the percentage of the time users can actually access a system. Availability is affected by the stability of the entire ecosystem. For Shift4, this would mean availability includes not only our uptime, but also the uptime of our merchant customers’ ISPs, processors, banks, etc. We work with one processor that goes down for an hour of maintenance every single night. For customers using that processor, availability is about 96%. Fortunately, we know which hour they’re going to be down each night, so we advise customers to only use that platform if they’re never going to process batches between 2 and 3 a.m. For a retailer that processes between 6 and 7 p.m. every night, the nightly downtime is irrelevant. However, for a 24/7 operation like a hotel or international website that always processes its batches in the middle of the night, it could be devastating.

Today, there are companies advertising five-nines availability, or 99.999% uptime. These companies claim less than 5.26 minutes per year of downtime, which works out to less than one second per day on average. Now, can a company go a whole day without a second of service interruption? Absolutely. Can they go a week? If their system is well designed, yes. But can they sustain that standard for a year or more? That is exponentially more difficult to guarantee.

Here’s how they get away with it. First, they write their contracts very carefully. The processor I just told you about with the hour of nightly maintenance doesn’t count that time as part of their advertised uptime. Their uptime is “net of scheduled maintenance.” Scheduled maintenance – essentially any downtime announced in advance – is written out of their service level agreement (SLA) contract, so it doesn’t count against their advertised percentage. Add it back in, and even if they never had a single minute of downtime outside of their hour window, they’d have only 95.83% uptime.

Another important thing to consider is what they guarantee versus what they advertise. Some are hazy on the difference between the two. Shift4 is careful to differentiate between what our SLA guarantees (99.7% uptime) and what we average (99.98+%). What’s funny about that average is that it’s cumulative for the last decade, but as our technology has steadily improved over that time, so has our uptime. In fact, DOLLARS ON THE NET has not had any measureable downtime since December 2009.

Let me say that again for the skeptics: in the last five years, we have recorded 100% uptime – a stat unrivaled by any other organization in the payments industry.

Why don’t we start guaranteeing that, or at least advertising it? Because that’s not the way Shift4 works. You can’t surpass five years with zero downtime without being incredibly cautious and meticulous. We’ve built our system on redundant servers, running in redundant data centers, with redundant power supplies. We’ve built adaptive routing technology into our Universal Transaction Gateway® and “taught” it how to find the best route from the merchant back to Shift4 – and how to avoid slowdowns and dropped connections. That being said, we’re not arrogant enough to believe we’re infallible. We know we’re good, and we’ll likely never drop below 99.99, but we don’t sell “likely,” we sell facts.

The facts are simple. Not one of the processors we interface to can match our uptime over the last five years. Their uptime is the limiting factor on the availability our merchant customers experience through DOLLARS ON THE NET. Our technology and infrastructure is meticulously designed and constantly monitored and improved upon as new capabilities become available.

What I’m saying is this: any company can claim five-nines uptime and can tell you they’re more reliable than Shift4 because we only guarantee 99.7 in our SLA. But the competition is not over who can write a sneakier contract or who can find the best way to skew their statistics. We’re confident in saying that no other payments company can match what we’ve actually provided to our merchant customers over the past five years – and what we fully expect to continue providing. We’ve had no merchant-impacting downtime in the last five years, and we expect our new data center at Switch SUPERNAP to make our solution even more reliable – although we haven’t figured out yet how you trump 100%.

Share |