Technologies : ASPs Push Service-Level Management Boundaries &mdash ASPnews.com

A compelling feature of the ASP sales pitch is the assertion that customers can rely on a consistent, world-class quality of service (QoS). ASPs are making those QoS promises in the form of service level agreements (SLAs) built into their client contracts. Their ability to live up to their promises depends on the service level management skills and technologies at their disposal.

Over the years, there have been numerous tactical approaches to network, systems and applications management, but management schemes have historically been piecemeal, often focussed on specific vendor products. While these tools are necessary tactical components, they do not answer the strategic needs of ASPs.

ASPs face the complex challenge of assembling and managing multi-provider SLAs. The components of an ASP solution may originate from separate internal divisions, or from an external supply chain that can include private and public network providers, infrastructure and data centre providers, systems integrators and technical support call centres, each with their own SLAs.

Each element must be separately monitored, measured and managed - but customers experience the resulting service as a single entity. They are not interested in the individual components, and care only whether the end result meets their expectations for availability and performance. So the ASP must have a strategy to combine and synthesise all the component services within a comprehensive SLA management scheme.

Weakest link

The end-to-end service delivery axiom - that you're only as strong as your weakest link - doesn't quite apply to ASPs. In fact, ASPs are not even as strong as their weakest link, because each service that offers less than 100% availability dilutes the overall effect. Total service availability is calculated by multiplying together the figures for each individual component, as shown here (Total Service Availability % = SA%1 * .* SA%N).

Network Provider
(WAN and internet facilities) 99.95%
Infrastructure Provider
(data center/systems uptime) x 99.50%
Application Management Services
(application/failover services) x 99.50%
Total Service Availability: 98.90%

Each individual component is guaranteed available for all but 50 minutes downtime a week (or just 5 minutes in the case of the 99.95% component). But if each fails at a different time, the overall service may be unavailable for almost 2 hours a week. Some providers recognise this by defining availability in terms of consecutive hours. So long as service is restored within 2 hours, even if it fails for a similar period on another occasion, it is considered 99.5% available.

Network SLAs

SLA monitoring is an established practice at the physical layer. For private networks, carriers report their own performance on metrics such as latency and availability, and offer service level guarantees even across multi-vendor private networks, having negotiated SLAs with other carriers. Some augment their core physical layer services with managed services, such as VPNs, managed router services or managed IP services, again delivered against SLAs. But a significant number of ASPs choose to manage these higher-level functions themselves.

There are various tools available to monitor and report on network performance. Routers from leading providers such as Cisco and Nortel Networks include management tools, while third parties offer additional tools to monitor performance of the Internet backbone. Specialist tools from vendors such as Visual Networks completely automate the collection, interpretation, and presentation of service level data.

Systems SLAs

There is a wide choice of tools available to monitor the systems environment in the data centre. They can be used to report on percentage uptime/downtime, number of outages, the longest and shortest outage, MTBF (mean time between failure), and MTTR (mean time to repair).

Many ASPs rely on the default tools for their server environment - for instance, those who deliver apps to Windows terminal clients typically rely upon Citrix Resource Services Manager and Microsoft Systems Management Server (SMS).

Server vendors such as Compaq, Sun and HP offer their own tools for performance management and availability. Additionally, in combination with or separately from default tools, some ASPs are building their performance reporting capabilities around higher-level third-party products.

Where ASPs outsource the data centre to a hosting provider, they must either manage the server farm remotely, or rely on service level reporting from the provider.