Enforcing SLAs with Real Data

I’m sure by now you’ve probably seen tons of articles telling you about how important it is to travel with location devices in your luggage. The most common one I’ve seen is the Apple AirTag. The logic goes that if you have one in your checked suitcase that you’ll know if there are any issues with your luggage getting lost right away because you’ll be notified as soon as you’re separated from it. The advice is sound if you’re someone that checks your bag frequently or has it lost on a regular basis.

The idea behind using technology to enforce an agreement is a great one. We make these agreements all the time, especially in networking. These service level agreements (SLAs) are the way we know we’re getting what we pay for. Take a leased line, for example. You typically pay for a certain speed and a certain amount of availability. The faster the link or the more available it is the more it costs. Any good consumer is going to want to be sure they’re paying for the right service. How can you verify you’re getting what you’re paying for?

For a long time this was very hard to do. Sure, you could run constant speed tests to check the bandwidth. However, the reliability was the part that was typically the more expensive thing to add to the service. Making a circuit more reliable means adding more resources on the provider side to ensure it stays up. Allocating those resources means someone needs to pay for them and that is usually on the customer side.

It’s easy to tell when a link goes down during working hours. You can see that you’re not getting traffic out of your network. But what about those times when the circuit isn’t being as heavily utilized? What about the middle of the night or holidays? How can you ensure that you’re getting the uptime you pay for?

Trust and Verify

This is actually one of the groundbreaking areas that SD-WAN pioneered for networking teams when it came out years ago. Because you have a more modern way of maintaining the network as well as a way to route application traffic based on more than source and destination address you can build in some fancy logic to determine the reliability of your circuits too.

One of the biggest reasons in the past to use a leased line over a broadband circuit was that reliability factor. You need MPLS because it’s more stable than a cable modem. You get guaranteed bandwidth. It’s way more reliable. Those are the claims that you might hear when you talk to the salesperson at your ISP. But do they really hold water? The issue for years is that you had no way of knowing because your analytics capabilities were rudimentary in most cases. You could monitor the link interface but that was usually from the central office and not the remote branch. And depending on the polling interval you could miss downtime events.

SD-WAN changed that because now you could put an intelligent device on the edge that constantly monitored the link for throughput and reliability. The first thought was to do this for the broadband links to see how congested and unreliable they could be do know when to switch traffic to the more stable link. Over time admins started putting those same monitors on the MPLS and leased line circuits as well. They found that while broadband, such as cable and DSL, was typically more reliable than the sales people would have you believe it was the relatively unreliable leased circuits that surprised everyone.

Much like the above example of lost luggage you previously had to take the airlines at their word when they said something was lost. No details were available. Did your bag ever leave the airport when you flew out? Did it make it to the right location but get stuck somewhere? Did your bag get on the wrong plane and end up in Cleveland instead of Las Vegas? Because the airline reporting systems were so opaque you had no idea. Now, with the advent of Tile and AirTags and many others, you can see exactly where it is at all times.

The same thing happened with SD-WAN analytics. Now, armed with proof that links weren’t as reliable as the SLA, admins could choose to do one of two things. They could either force the provider to honor the agreement and provide better service to meet the agreement. Or the company could reduce their contract to the level of service they are actually getting instead of the one that is promised by not delivered. Having the data and a way to prove the reliability helped with the negotiations.

Once the providers knew that organizations had the ability to verify things they also had to up their game by including ways to monitor their own performance to ensure that they were meeting their own metrics. Overall the situation led to better results because having technology to enforce agreements makes all sides aware of what’s going on.


Tom’s Take

I’ve only had a couple of bags misplaced in my travel time so I haven’t yet had the need to go down the AirTag route. I’ve done it on occasion for international travel because knowing when my bag is about to come out on the carousel helps me figure out how much time it’s going to take me to get through customs. The idea of using technology to make things more transparent is important to me on a bigger scale though. If you can’t verify the promises you make then you shouldn’t make them. Having the AirTag or SD-WAN software to ensure we’re getting what we pay for are just parts of a bigger opportunity to provide better experiences.