How Granularity Exposes the Truth About SLAs

True story.

A business is having intermittent issues with some of its cloud applications. They aren’t performing consistently. So the business talks with its application vendor, and the vendor says everything is smooth on their part of the network. It isn’t them.

So next, the business reaches out to its network service provider. The service provider does some tests, and consults a few dashboards. Again, things look good on the service provider level; according to their data, all SLAs are being fulfilled.

There’s still an intermittent problem, though, so the business talks with various parts of its IT team. Everyone says it isn’t them, and things look good. It really must be the service provider, but they have no way of proving it, because all the tools they have show green. Yet, the end users are still complaining.

So this business approaches Accedian for help. What’s going on? There’s a problem, but nobody can figure out what’s wrong with the network, and the applications are still not performing consistently. 

That’s where Accedian comes in to help. Accedian uses its cloud-based, real-time Skylight network and application performance monitoring solution to take a look at the actual dynamics on the network. No surprise, there’s lots of red. The service provider network is not delivering, which is what the business originally suspected.

Of course, the service provider cries foul and says that’s impossible; their internal tools show that the network is fine and all SLAs are being met. So then Accedian shows the service provider that its tools are not measuring precisely enough. There’s not enough granularity in its network tools. There really is a problem and a breach of the provider’s SLA, but the problem was hiding in tools that weren’t able to see to the level of detail needed

Why measuring the network in microseconds matters 

Measurements can be misleading if you are not measuring at the right levels of granularity. Every microsecond matters in terms of quality of service and experience. 

Sprinters in the Olympics used to be measured by a stopwatch. For a time, that was enough to know who won or lost a race. But as athletes improved, and sprinters got closer to each other at the finish line, the seconds on a stopwatch stopped capturing the actual measurement needed to determine who won the race. More than one runner finished at the same second, so milliseconds were then needed to gauge who actually won a race.

The same situation applies to network performance in 2023. Without granularity that monitors an organization’s network performance at the sub-second level, important dynamics on a network can get missed. Dynamics that determine if an application runs consistently or experiences periodic network trouble. Dynamics that reveal if a network service provider actually is meeting its SLAs or only thinks it is.

Network performance only looked good when averaged

The network performance problem with the client business mentioned above came down to measurement granularity.

The business in question relied on a series of bursty applications that would send lots of data at the same time, then almost no data the next second. There would be traffic spikes that would tax the network and exceed allocated capacity, causing application performance degradations. But these spikes would only happen for a second or a few milliseconds. The rest of the time, there was plenty of network capacity.

To the service provider’s measurement tools, the network looked good. The true data was hidden, when network usage was averaged, it looked like there were no network performance issues. But when Skylight measured network performance using more granularity, looking at traffic at the sub-second level, the issues with network capacity became clear. That’s why the service provider thought that it was meeting its SLAs when really it was not. It wasn’t measuring precisely enough.

The need for granular network and application performance data

Not all business applications need precise performance. If a network sensor is polling once every 20 minutes, or a spreadsheet is uploading to a server, microbursts that impact performance and less precise measurements are not big problems.

For many applications, however, precise network performance matters. For a mechanical arm on the shop floor, it matters. For a financial transaction sent to a trading floor, it matters. For a business in the middle of digital transformation that is baselining and allocating network capacity so applications run smoothly, it matters. For a firm that is trying to prove that a service provider is not meeting an SLA, it matters.

With businesses increasingly relying on connected devices that work in unison and more complex network architectures, 15 or even 5-minute polling and averaged measurements no longer are enough. Network and application performance tools need granular visibility to ensure performance and more easily help spot issues when they arise.

This granular, end-to-end, real-time visibility of an organization’s network is at the foundation of Accedian’s Skylight technology. Without full network and performance visibility, down to the microsecond, it can be hard to identify emerging network issues or solve difficulties in the network that have already arisen.

For a look of how Accedian Skylight can help your business with network and application performance (or to enforce a broken SLA), schedule a demo today.