For decades, we’ve been conditioned to believe that fiber internet is something of a silver bullet. When a business has internet performance issues, they’ll commonly get the question, “What do you expect, you don’t have fiber?” So, it’s no surprise that folks ask us quite often whether Bigleaf customers still use our service after moving to fiber.
Spoilers: The answer is a big “yes.” In fact, 39% of the circuits currently used by Bigleaf customers are fiber. But since we monitor all our customers’ internet circuits 10x/second for outages and performance degradation, we thought we’d dig into all that data in aggregate to see exactly how reliable fiber internet really is and how much Bigleaf can actually help.
For this analysis, we’re focusing on fiber internet performance relative to performance-sensitive cloud applications like VoIP or UCaaS. Specifically, we looked at every fiber connection that Bigleaf monitored in the month of September and took the mean (filtered for outliers) of each of the following metrics:
- Downtime (mins) – The connection is completely unavailable.
- Unusable uptime (mins) – The connection performance (packet loss, jitter and latency) is so bad that something like a VoIP call would be unable to maintain a connection.
- Degraded uptime (mins) – The connection performance is bad enough that performance sensitive applications would be affected (e.g. calls would be choppy, echoey or robotic).
- QoS Adjustments – Number of times that Bigleaf algorithms had to reduce effective circuit throughput to because the carrier-rated circuit speeds were not being delivered fully and cleanly. These adaptations were required to ensure sensitive traffic was prioritized, providing proper application performance.
The resulting data was a little surprising, even to us. Here’s what we learned.
Fiber makes up almost 40% of the circuits used by Bigleaf customers
This wasn’t surprising to us as fiber is a wildly popular option for business internet. Many of our customers will use fiber with a less expensive Cable or DSL line that can be utilized for commodity bandwidth and failover.
That begged the question, for those with fiber lines, how often was a second circuit needed to handle failover. Turns out that…
Fiber connections had over an hour of downtime per month on average
This didn’t seem right, so we checked the data twice. Across all our customers’ fiber lines, even after filtering out those that were down for days at a time, there was an average of 62.92 minutes of outage time during the month of September.
It’s important to note that those 62.92 minutes were very likely in small increments of a few seconds or minutes. If someone was downloading a file during an outage like that, it might resume eventually. But if a sales rep was on a VoIP call during one of those outages, the call would drop. If we assume an average of 10 seconds per outage, that would mean that a company using VoIP over fiber risks as many as 372 dropped calls per month.
But at least fiber circuits are reliable when they’re up, right? Well…
Fiber connections were unusable for an additional 50+ minutes per month
Beyond the hour of total downtime, the average fiber line experienced another 50.6 minutes per month where the circuit was live, but performance (packet loss, jitter, and latency) were so bad that a performance-sensitive application like VoIP wouldn’t be able to run.
This is almost as bad as having a dead circuit, but with one added challenge… if you’re using a dual-wan firewall for failover, it would still see this circuit as live and it wouldn’t fail over. This means that, in order to get your calls and applications back up and running, you’d either need to manually change to another circuit or use an SD-WAN like Bigleaf with a second circuit to fail over seamlessly.
So we’re up to almost 2 hours of application downtime per month on the average fiber circuit. What else could go wrong? Sooo…
Fiber connections experience almost 9 hours of degraded performance each month
Beyond being effectively down for an hour and 50 minutes per month, we measured an additional 531 minutes (about 8.9 hours) of time when packet loss, jitter, latency and throughput were bad enough to affect application quality.
For someone who relies on a VoIP phone to do their job, that would translate to 8.9 hours of choppy or echo-y calls. Not a great look for their company.
The situation would be even worse if Bigleaf wasn’t able to correct for circuits where the rated speeds don’t match the actual speeds…
Bigleaf performed an average of 222 QoS speed adjustments on fiber circuits
Bigleaf automatically detects and prioritizes performance-sensitive traffic like VoIP, UCaaS, CCaaS, remote desktop, etc.. For any application prioritization to work, network devices must know how fast the circuits are that traffic is flowing through. Bigleaf’s patent-pending Dynamic QoS detects variance in circuit speeds in real-time, and adjust automatically, ensuring real QoS control over the highly-variable internet. Each time our algorithms detect that the circuit isn’t delivering the rated speed, our systems perform what we call a QoS adjustment to adapt to that and protect traffic.
Since we’re up to an average of over 10 hours of downtime or degraded performance per month over fiber, it’s no surprise that Bigleaf is making an average 222 QoS adjustments per month on those same fiber lines. Those connections simply don’t deliver their rated bandwidth at all times.
In other words…
Moving to fiber internet isn’t a silver bullet for your internet performance issues
That’s not to say that fiber won’t be an improvement over your existing ISP. Clearly, a lot of Bigleaf customers use and stick with fiber as their primary internet circuit.
But if your goal is to guarantee uptime and performance for your company’s applications, fiber may be a let-down. There’s no one perfect internet circuit type currently available. In fact, our data shows that fiber is really no better than several other circuit types when it comes to outages. That’s why so many businesses rely on Bigleaf to leverage multiple circuits to create a dynamic, intelligent, active/active connection to the applications they rely on.
We hope this data was helpful. If you have any other internet performance-related questions that we can answer for you, let us know and we may use it for a future blog post. If you’d like to see if Bigleaf can help you with your own internet outages and performance challenges, contact us today for a free consultation or free 30-day trial.