What Is a Service Level Agreement? Everything Businesses Need to Know About Internet SLA’s
Organizations that rely on broadband connectivity for business operations need to understand how an ISP’s service will perform day in and day out. A Service Level Agreement (SLA) details the performance parameters of the service. It also provides the remedies, often in terms of service credits, that a customer will be due if the services fails to meet the minimum performance specs.
Because the SLA guarantees a stated level of service quality, it’s a key tool in determining if an internet service is right for your network requirements. Reviewing SLAs prior to signing up for service will not only help you analyze and compare the expected service quality across providers but also select the best services for your business.
The core components of SLAs are uptime, packet delivery and latency.
What Is Uptime in an SLA?
Uptime is defined as the percentage of time that a customer has broadband service availability in a given month.
Business grade services that are backed by SLAs, such as fiber internet service or fixed wireless internet strive to achieve 99.99 percent or better service connectivity. This means that the internet service will support an average downtime of about 4 minutes or less per month. In comparison, a service with 99.9 percent uptime, which sounds reliable until you do the math, may experience 44 minutes per month of downtime.
Once you understand the impact of uptime, you’ll want to learn how the provider defines it. Most ISPs include their core network and all the way to the customer’s premise (DMARC) in uptime metrics. But some SLAs will limit the service guarantee only to the carrier’s core network, excluding the customer access network. Uptime for only the core network has minimal meaning for customers since it eliminates any guarantee for the last leg of the connection that links directly to your office.
Additionally, keep in mind that internet providers resell services that are delivered over other carriers’ networks. These are known as a type II circuits. Type II circuits are often covered by an SLA that is separate from services provided directly over the carrier’s own network. Before you sign, understand which internet service and network the sales rep has quoted so you know which SLA criteria apply.
What Is Packet Delivery?
Packet delivery is defined as the percentage of data packets received with respect to data packets sent. It can also be understood in terms of packet loss (the percentage of packets not delivered with respect to packets sent). The standard metric for packet delivery for business-grade services is 99.5 percent. If packet delivery falls significantly, you may experience much higher latency.
What Is Latency?
The amount of time it takes a packet to travel on the network is known as latency. It’s most often expressed in milliseconds (ms). Latency metrics may be given for the core network alone or for the core network plus the customer access network. SLAs may also state the metrics as one-way latency or roundtrip latency. Pay close attention to the definitions to accurately assess and compare the latency components of SLAs.
Latency is most important to businesses running real-time applications like video conferencing, surveillance systems, or VoIP. One-way latency of 100 ms or less is well-suited for these applications. But if you’re a competitive online gamer with fast reflexes, then 50 ms or less is ideal. Studies show that even the most attentive eyes can’t perceive latency under 20 ms. So small differences in latency, less than 20 ms, don’t meaningfully differentiate services. Commercial-grade services typically guarantee a one-way latency of 25 ms or less, which is suitable for nearly all cloud apps.
Do I Need MTTR?
Some carriers also add MTTR to their SLAs. MTTR refers to “Mean Time To Respond” or “Mean Time To Repair.” Mean Time To Respond is generally used to convey the average maximum amount of time for a carrier to respond to customer inquiries regarding an outage or severe degradation issue.
The response can be in the form of running diagnostics, scheduling a tech visit, and providing status updates to customers. It doesn’t include the Time To Repair the issue. A typical Mean Time To Respond metric is 4 hours, on average. It usually applies to business hours only, excluding holidays. Again, this is the “average” time to respond – a monthly average across all customers. As an individual customer, you are not guaranteed a four-hour response time.
Mean Time to Repair, on the other hand, is the maximum time per month, on average, to repair an outage or severe service degradation issue. It may or may not be limited to business hours or exclude holidays.
While many SLAs don’t include an MTTR, the absence of MTTR alone isn’t a reason to be concerned. MTTR directly impacts uptime. If the carrier is slow in fixing an outage or severe service degradation, then the service availability is negatively affected. Hence, carriers are motivated to repair outages and severe service degradations within uptime metrics even when no MTTR is provided.
Service credits are the typical compensation given for missed SLA metrics. The service credits can be cumbersome to compare. A case in point is the promise of 100 percent uptime. We all know that every service and product has downtime. Nothing is 100 percent. Despite this, some SLAs tout 100 percent uptime. But if you read the fine print, the remedies don’t apply until the service has dropped to 99.95 percent, which can be the same level at which uptime guarantees of 99.99 percent. For chronic issues, contracts and fees can usually be terminated without penalties. The astute buyer will dig down into the service credits to fully assess the predicted level of performance for their business as well.
Does Everyone Need a SLA?
Eighty-five percent of businesses are using cloud-based apps today. For most of these businesses, a high level of service quality and connectivity is required. The caliber of service you need will ultimately depend on your network requirements.