The devil's in the details when comparing service-level agreements

     Have you noticed the rash of guarantees these days?
     From sub sandwiches to convenience store gasoline to data network 
services, companies everywhere are anxious to show you how much they 
believe in the quality of their products and services.
     In the carrier world, things are no different. Service-level 
agreements (SLA) are becoming the rule rather than the exception.
     If you haven't signed an SLA yet, one likely will be in your future. 
But taking advantage of everything an SLA has to offer, in terms of reining 
in costs and guaranteeing response times, can be a tricky situation.
     Negotiating an SLA isn't as simple as playing one vendor off another 
because the real differences aren't in the numbers but in the methodology 
for generating the numbers.
     Comparing one provider's idea of a guarantee with another's is like 
comparing apples to armpits.
     There is no standard for measuring availability, throughput or latency.
     Some vendors calculate availability, for example, based on the time of 
open trouble tickets.
     If the network goes down in the middle of the night and you don't get 
your trouble ticket until the morning, you can't count the actual downtime 
against the SLA.
     Others offer network availability guarantees but not end to end. So if 
the loop goes down and stays that way, you still won't get a credit.
     And what about the credits?
     MCI gives back between 5% and 15% of the total network monthly 
recurring charges if it doesn't deliver on the guarantee in any month.
     But like many carriers, MCI measures the availability networkwide.
     So for a five-node star network, you could have a remote site down for 
a full day and still be at 99.7% availability for the whole network.
     For a 25-site star network, a remote location could go down for three 
days and you'd be above 99.5%.
     In either case, only if the primary site went down for four hours 
would you drop below the 99.5% threshold.
     Network availability guarantees can be tricky, too.
     Take frame relay services as an example. Network availability is 
guaranteed at 99.5% at MCI, 99.9% at LCI and 99.99% at Sprint. Sounds 
pretty good.
     But is this end to end or point of presence to point of presence? 
Sprint, for example, gives its guarantee POP to POP and gives an end-to-end 
guarantee of 99.5%. 
     Look into how availability is calculated when you're comparing across 
companies, not just at the numerical value of the guarantee.
     For larger networks, it might be more valuable to negotiate a 
per-location availability guarantee.
     As our friend Tom Nolle, president of the CIMI Corp. consultancy, 
sometimes says: It doesn't matter if the carrier sprinkles fairy dust over 
your network and delivers your data to its destination on Tinker Bell's 
little wings, just as long as the performance lives up to the promise.
     By the way, we'd like your stories on successful SLA negotiations. 
What did you win, and how did you do it?
     We plan to compile the more interesting and insightful information and 
pass it along in a future column (names will be changed, where requested, 
to protect the guilty).
     Briere is president and Heckart is vice president of TeleChoice, 
Inc.,a consultancy in Verona, N.J. They can be reached at dbriere@ 
telechoice.com and checkart@ telechoice.com.