The Changing Face of Service Level Agreements

The cable industry is trying to sell its commercial services to bigger organizations. The long-term goal is to be a player among multinational, Fortune 500-type organizations. But first things first: At this point, a great deal of the action is to more actively romance medium-size businesses that are a bit bigger than the small offices/home offices (SOHOs) and very small businesses that it has focused on to this point.One thing that the sort-of-big guys have in common with the bona fide giants is the demand for service level agreements (SLAs). In other words, truly small companies likely are served by the residential HFC plant, perhaps with some special features. Bigger organizations are more apt to take advantage of fiber and the SLAs that accompany it.There are two types of SLAs: Those aimed at performance and those aimed at operations. The operational SLAs focus on the actions of the service providers: Is the service provider instituting new or amending existing services in the agreed-upon timeframe? What is the mean time to repair?The other SLA deals with how the network is operating and, to date, has focused on parameters such as jitter and latency. In general, cable operators are holding their own in the world of SLAs. "Cable operators, because they have moved up market, now are competing head to head for mid-size business and have created SLAs that pretty much track with those that are offered by the incumbents," said Barry Zipp, industry marketing director for Ciena.Big changes are coming to performance-based SLAs. Those changes will impact telephone companies and other providers as acutely as cable operators, experts say.The Open Systems Interconnection (OSI) model is a seven-layer cake of protocols that govern how networks link to each other and trade data in a way that is mutually usable. The lower levels deal with establishing connections. The higher the level, the more sophisticated the task. At layer 1, for instance, is the physical layer (Is it fiber network? A wireless connection?). Layer 7 deals with actual applications (How, for instance, an instant message is sent in a way that can be displayed by the receiving network).Today's SLAs mostly focus on layers 2 and 3 of the OSI protocol stack. The problem is that many of the applications businesses rely upon today rely on transmission control protocol/Internet protocol (TCP/IP), which makes demands that must be satisfied at higher levels. In other words, it is possible that a service provider could meet an SLA based on layer 2 and 3 while, at the same time, the applications that a company relies upon are not working because of a problem at a higher level in the protocol stack, said Rachel Weiss, a senior product manager at JDSU.The key reason that a session may be fine at the lower level and pass muster according to currently used SLA measures but fail at the higher levels is that TCP/IP-based applications are "bursty" in nature. As the name implies, bursty applications tend to send data in dense waves. This, Weiss said, can "stress the network in ways that layer 2 and 3 cannot and expose misconfigurations that aren't revealed at layer 2 and 3."On one level, the answer simply is to include tests for bursty traffic in SLAs. And, indeed, such testing protocols are emerging. Weiss pointed to IETF 6349, a three-year-old test that can measure busty traffic at layer 4.Two other protocols MSOs are working with are the IETF’s 2544 and ITU-T Y.1564, said Ricardo Torres, the director of product marketing for VeEX, which completed its acquisition of Sunrise Telecom last month. The protocol soup is confusing in that the protocols don't measure precisely the same attributes. The bottom line, though, is that they all are geared toward helping test networks as they operate today.The real challenge isn't the technical ability to make these networks work or the creation of test suites capable of measuring their success. It is divvying up responsibility. Currently, Weiss and Torres said, businesses are doing their own testing to provide evidence that the SLAs are failing or, perhaps more accurately, that SLA compliance is basically meaningless if the application isn’t working.Service providers, on the other hand, are reluctant to move the focus to the higher level testing because they feel that the problem spots that keep applications from working are on the customer's LAN and therefore not their responsibility. In other words, if latency, jitter and other layer 2 and 3 measures are in order, the service providers feel that they have successfully done their jobs.There is a lot happening in the performance SLA sector. The good news for cable operators is that these issues are being worked through before the focus turns fully to enterprise SLAs.Carl Weinschenk is the Senior Editor of Broadband Technology Report. Reach him at [email protected].