Room for SLA improvement - Serving the Corporate Market

The lack of 'bite' to traditional SLAs has opened the door for third parties to negotiate more meaningful service level agreements between carriers and their business customers.

SLAs between carriers and IT managers have, all too often, been pretty meaningless. Why? Because, invariably, they have been drafted from the perspective of the carrier -- not the end-customer -- and are therefore unrelated to the real needs of the business. Moreover, the penalties for non-compliance have often been insufficient to encourage carriers to address these issues.

Given that service level agreements are a critical feature of the IT landscape -- particularly in the managed data services sector where strict guarantees around network performance have to be in place -- IT managers are increasingly turning towards outsourcing partners to negotiate more meaningful SLAs. According to the Yankee Group, the process of transferring business-critical yet non-core services to trusted third parties has moved into the mainstream over the past five to ten years and is now estimated to be a US$95 bn industry.

A reason for this emerging trend is that 'traditional' SLAs are of the vanilla variety only -- 'one-size-fits-all' metrics are applied by carriers to all customers with little in the way of customisation. SLAs of this sort also tend to play to the carriers' core strengths (resiliency of the core backbone network with availability -- and perhaps latency -- being the key components). Although having the network up and running is undoubtedly crucial, there are some more detailed criteria that need to be considered.

A different approach

If SLAs are to have more 'bite', SLA metrics clearly need to be monitored independently. If a carrier monitors them internally then it's the equivalent of putting a fox in charge of the henhouse. Having said that, there is a growing trend towards providing network performance statistics online and in real-time. This provides customers with some means of confirming the information they are receiving from the carrier and is an essential tool if a carrier's adherence to the agreed SLAs are to be monitored effectively.

The second point is that availability and latency, while important, are blunt instruments when it comes to the overall performance of the network. The following is a variety of other metrics that may have a stronger bearing on ensuring that mission-critical business processes are maintained:

* Proactive versus reactive fault resolution -- carriers who manage the network on behalf of the customer have the best overall visibility of its performance. They are therefore best able to identify any problems and, in an ideal world, carriers should be able to commit to detecting any problems before they even become visible to the customer. Pro-active identification of 70 per cent of faults may sound unreasonably optimistic, but customers can obtain these kinds of commitments from some network providers. Taking these kind of metrics into account at the SLA stage puts the onus of fault identification onto the partner provider and can have a hugely beneficial impact on the performance of the network.

* Time to fix (not time to site) -- this is not simply about fielding a junior employee who may or may not have the experience to resolve a problem within a fixed time simply in order to tick a box and keep a promise made to a customer. (Wired Magazine in the US refers to these employees as 'appeasement engineers'.) Having 80 per cent of problems fixed within five minutes and resolving 98 per cent of faults without a site visit is a realistic commitment. The goal should be problem resolution, not simply a meaningless olive branch.

* Time to provision -- the transition from an in-house network to one provided by a third party can be complex. The last thing an organisation needs is for delays in provisioning at the start of the relationship to impact on the performance of the network. Also, networks are dynamic organisms that need to be fine-tuned as business requirements change. It is important that meaningful timescales for the implementation of the core network and any subsequent upgrades are written into the SLA.

* Management reporting -- online reporting is becoming more important as a means of providing visibility of network performance. However, not every stakeholder in the outsourcing process has the time or the inclination to look through (or understand) such information. Have you established the level, regularity and detail that are to be provided as part of the reporting mechanism?

* Application-specific SLAs -- nobody wants a network for its own sake; it is there as an infrastructure for business applications. After all, the network manager of a transportation company may be able to live with intermittent interruptions to e-mail traffic but will take a very different stance to the same problems occurring with the company's core logistics package. At the moment, I know of nobody able to provide application-specific SLAs, but these types of guarantees should be available over the next year or so. Again, this is a conversation that customers should be having with their network provider.

Another key issue is whether SLAs are truly end-to-end. n many cases, performance guarantees are only applied across the carrier backbone while the tail circuits -- the 'last mile' of connectivity from the carrier's PoP to the desktop -- are often excluded from any agreement. This can lead to a finger-pointing exercise with the carrier saying 'it's not our fault' -- scant consolation to the end-user who is denied access to applications or network resources. In truth, this is beginning to be addressed by some carriers, but the penalties for non-compliance are usually laughably inadequate.

SLA negotiation hurdles

The area of SLA negotiation is a difficult one for both enterprise customers and carriers to address. Enterprises, even large MNCs, simply don't have the bandwidth requirements that will give them the 'clout' to negotiate meaningful SLAs with carriers. In any case, this kind of complex negotiation is far from their core business and, frankly, beyond the resources of most. It is therefore logical for them to turn to third parties to have these kind of guarantees put in place.

Carriers are in a much different position. Because of the huge costs involved in building out their core network infrastructure, carriers are disinclined -- to put it mildly -- to identify and work with third parties that may provide a more resilient or more cost-effective solution in a particular geography. (We refer to this as MPLI -- Must Protect Large Investment -- syndrome.)

Lacking experience in negotiating with third parties renders the integration of third parties into SLAs particularly difficult. In any case, embracing this approach moves the carrier much further down the virtual, 'asset-light' approach than they are prepared to countenance. Building and maintaining relationships with third parties as a core competence, not as a necessary evil, is the only way to ensure real performance guarantees across the entire network.

So, the first step in negotiating SLAs is to make sure that the criteria genuinely reflect the requirements of the business. For example, a global law firm which creates and amends documents in real-time across the network will naturally place a higher priority on latency guarantees. For a high street retailer reliant upon credit card transaction details being transferred, the latency aspect of performance will be less critical than say availability. The second step is ensuring that the penalties for non-compliance have real teeth and provide more than just a slap on the wrist for an offending carrier.

Typically, penalties for non-compliance (assuming that the end-customer has the visibility of the problems in the first place) are usually a rebate of a few days service time. This is a pin-prick for most carriers and is particularly true of the penalties associated with the non-performance of the tail circuits -- they are simply insufficient to merit attention and resolution. An SLA that strikes at the very lifeblood of a carrier, would provide a real incentive for change.

When Vanco strikes an SLA with its customer, we contract to rebate a significant proportion of our management fee if we fail to satisfy the criteria laid out up front. As this is our primary source of income, non-compliance has a direct impact on our bottom line and we are therefore highly motivated to ensure that problems either don't arise in the first place or are fixed promptly.

If there are serious, ongoing problems with the performance of a network, or with significant portions of a global network, then customers should have negotiated the right to swap out that part of the network for an alternative. This is something that carriers will be very unwilling to provide since their business model relies on the economies of scale arising from customers using the network on an end-to-end basis. Nevertheless, if this is not an option open to a customer, then the value of the SLA must be called into question.

In its most extreme form, significant network underperformance should be punishable by the cancellation of the contract -- without penalty. Naturally, the terms under which this can be invoked need to be carefully agreed, but that is true of any element of an SLA. Again, customers trying to negotiate this with a carrier should expect extreme pushback as the carrier business model relies on heavy up-front investment recouped over a long period of time. But if a network is a mission-critical part of a business' infrastructure then this must be an option if the customer is to have any peace of mind.

As with all things, insight comes with experience. I think it is fair to say that to date, the drafting of SLAs has been heavily weighted in favour of the carrier. As a result, their customers have been dealt a pretty poor hand when it comes to enforcing meaningful penalties for network performance. However, the fact is that SLAs can be structured to provide a high level of performance guarantees and appropriate penalties if these are not met. If there is an upside to the current telecoms crisis then it is that carriers should be much more highly focused on servicing their customers than they have perhaps been in the past. Similarly, in what is now an incredibly competitive telecoms environment, customers are now better placed than ever to extract from carriers service levels that have a meaningful relationship to their business. If we can make the economic realities bite sufficiently hard then we can emerge with the telecoms industry we deserve, not the one that we are saddled with at the moment.

Andy Summer, account director, Vanco

COPYRIGHT 2002 Horizon House Publications, Inc.
COPYRIGHT 2003 Gale Group