The service level agreement (SLA) has long been a sore spot with enterprises. Its legalese can be intimidating and confusing, and its terms can at times almost appear too good to be true. So what do you need to know before you sign on the dotted line?
The SLA is a written guarantee of what the provider will be offering you, and indicates what penalties or compensation will be provided if they falter. Ultimately, it is your protection. Depending on the type of service you are receiving and the provider offering it, the SLA will vary. For instance, the issues covered in an SLA from a collocation provider or hosting service will differ from that of an SLA from a network services provider.
While it might be easy to take a defensive approach when dealing with your provider during the original negotiations of your SLA, that should not be the case.
“I think the key thing is that SLAs need to adequately represent the capabilities on both sides. They have to be clearly understood, and it is not an ‘us vs. them’ situation,” said Mark Fabbi, Toronto-based vice-president and research director with Gartner Inc. “It really should be dealt with as a partnership. You want an outsourcing situation to succeed. You are not going into it because you are going to war; you’re trying to make your infrastructure or systems better.”
In fact, Fabbi stressed that an important thing to do before the preliminary negotiations of an SLA is to work through possible scenarios.
“Don’t trust the value that is on the piece of paper,” he warned. “You actually have to work through and really understand what that means if things start going wrong, and in any sort of service case, things are never going to be perfect.”
In particular, he said to keep in mind what will be considered mission critical in your situation. It is important to identify things that are a little out of the ordinary, and to identify the hot spots that need special treatment or priority should something go awry. What are the specific service levels your company requires?
“That needs to be discussed up front and documented,” said Fabbi.
Tim Gilbert is the chief technology officer at research firm Morningstar Canada Inc. in Toronto. The company produces research for the mutual fund industry, and caters primarily to advisors, private investors and the mutual fund companies themselves. With the research that it creates, Morningstar also produces software products, which it sells to those industries, Gilbert said, adding that one of its products is its Web site. The company engaged Q9 Networks Inc. in Toronto to house its equipment – which meant an SLA.
Now a seasoned veteran when it comes to SLAs, Gilbert started out knowing nothing about them. It was when the time came to provide them to Morningstar’s clients, he said, that he had to learn fairly quickly. Gilbert found templates from various ISPs, and was able to pick and choose elements he liked from those.
“I had to understand what I could afford to promise to our clients,” he explained. “And that is, in our case, very dependent on what the service that our ISP provides us, so it is sort of like a domino effect. If our ISP is down, our clients are down, and that affects our SLA…So in writing our own SLA, we had to go through basically every sentence and figure out what all this meant.”
In the case of Q9, all of its customers get an SLA when they sign on for services, according to the company’s CEO, Osama Arafat. He noted that Q9’s SLA is pretty much standard from one customer to the next.
“But as far as Q9 is concerned, we will generally listen to the customer’s concerns, and see if there is an area we can improve it.…We’ll try to incorporate that improvement for all our customers.”
At Teleias Inc., a remote network monitoring and management firm, an SLA is based on availability and performance, depending on what the customer wants to do with his or her business, according to the company’s chief operating officer, Coleman Washington.
“They want to work with vendors that can adapt to their business needs,” Washington said. “There’s the old adage form Henry Ford: ‘You can have this Model T in any colour, as long as it is black.’ Well, the adage is that people want other than black, and more and more vendors are starting to evolve where they understand the business needs of their customers.”
There are certain things that are usually always included in an SLA, and you should be looking to make sure they are written to your satisfaction. Uptime is a measure of the percentage of the time the service provider guarantees to be online and up and running. Nowadays that percentage range is usually between 95 and 99.999 per cent, although some even guarantee 100 per cent.
“One of the things that is an easy mistake for people to make is when they look at the uptime guarantee, they take it as, ‘Okay, it’s going to be up 99 per cent of the time, or 99.999’.…But what they fail to consider is how generic network maintenance is,” Gilbert noted. “Network maintenance is almost never included in the uptime level guarantees, and depending on how general the network maintenance is, they could slip through a lot of downtime and still maintain their SLA.”
Another thing to look for is what is known as an escalation process or procedure, which identifies to what level in the service provider organization issues will be escalated should something go wrong. Using this procedure means that there is someone at the first level to call should you experience downtime. If something does not happen within a certain timeframe – say 15 to 30 minutes – the escalation procedure is followed and you go up one director or executive level to the next person on the list.
“At some point you want to hit somebody that is senior,” Gilbert advised. “You want to hit a director or vice-president. In our SLA, our highest level of escalation is actually the chief operating officer of the company, so we go all the way up. And that’s important, just as if you are not getting service, you need to know to make sure that you are going to get the attention that you deserve.”
Gartner’s Fabbi also warned that any service level has to be measurable by both sides. A service level on which only the service provider can measure, understand and report doesn’t create a lot of value, he noted, and one that is not measurable at all is even worse.
Fabbi also recommends a scheduled and ongoing review of how things are going.
“Even if a month has gone by and nothing really significant happened, you still have to go through some kind of operational review,” he said. “It helps with the ongoing relationship between the two companies, and keeps the ties much closer than if you just react to problems.”
