Tim Armandpour, VP, Engineering, PagerDuty, gives his take on how IT leaders can negotiate good service level agreements with vendors.
A number of companies are now using a range of cloud applications. Their uptime performance is measured by Service Level Agreements (SLAs). These agreements define different expectations and responsibilities, which should be met by vendors during instances of glitches, system crashes and downtime that may impact an organisation’s business continuity.
Gartner estimates that downtime caused by these issues can cost major corporations as much as $100,000 per hour.
However, not all SLAs are created equal, and worse, some SLAs are downright deceptive. Therefore, IT decision makers must evaluate SLAs to assure they cover minimum requirements for business continuity. When negotiating SLAs, watch out for these common traps:
SLAs with matching numbers of 9s are equal. If a vendor’s SLA guarantees a certain percentage of uptime during the month — such as 99.999 percent, or ‘five 9s’ — that means the site won’t be down more than one percent of the time during the month, right?
Don’t be so sure. First, get in writing that the definition of downtime includes not just ‘inaccessible,’ but also considerations for unreliable and unusable service caused by service performance degradation.
Another common malpractice is when vendors exclude scheduled maintenance from their definitions of downtime. Don’t let them get away with it. Customers will be upset any time they cannot transact necessary business, whether one of your vendors foresaw the problem and ‘scheduled’ the downtime or not. You don’t want diminished customer turnover because your vendor allows timeouts that don’t count against their five 9s.
Look for vendors who have eliminated scheduled maintenance by using iterative maintenance strategies that avoid downtime. If your vendor doesn’t use this method, be sure to clarify the following areas:
- Performance indicators (What does ‘so slow as to be unusable’ mean?)
- Problem management (What problem resolution responsibilities fall to the customer?)
- Terms and penalties (If terms aren’t satisfied, what exactly are the penalties and how are they enforced?)
‘Satisfaction guaranteed or your money back’ is a panacea. It turns out even the most generous fee refund or penalty for downtime doesn’t cover your losses. Here’s why:
Consider a cloud vendor’s self-described ‘industry-leading SLA’ of 99.95 percent that credits your account five percent of the paid-in-advance monthly fee for each 30 minutes of downtime, up to 100 percent of your fee. Full refunds are only issued when downtime reaches 10 hours in a month. More importantly, a fee refund is trivial compared to the true potential cost of a 10-hour outage, which could include a loss in sales and customer confidence and result in customer churn.
SLAs will scale with my business. SLAs are negotiated to meet the needs and size of the customer at the time of negotiation. Many businesses, however, change dramatically in size — even as often as quarter over quarter. A good SLA will outline specific intervals for reviewing the contract to ensure it meets changing business needs.
Some vendors take this a step further by building in notification workflows that indicate when an SLA is close to being breached so you can begin negotiations based on changes in scale. These tools also help ensure that scheduled SLA review meetings actually take place.
Good SLAs protect both businesses and suppliers from disappointment and missed expectations. When entering SLA negotiations, IT leaders can protect their business interests by clarifying what five 9s entails, by seeking vendors with methodologies to avoid downtime and by setting specific intervals for contract renegotiations that can scale with the business.
