THE INCREASING ROLE OF SERVICE LEVEL AGREEMENTS

IN B2B SYSTEMS

Philipp Masche, Paul Mckee, Bryce Mitchell

Research & Venturing, British Telecommunications plc., Adastral Park, Martlesham Heath, Ipswich IP5 3RE, UK

Keywords: e-Business, B2B, SLA, GRID.

Abstract: Service orientated architectures (SOA will support a dynamic market in commodity services and enable

business to drive down costs and respond faster and flexibly to changing markets. Virtualisation delivers

similar benefits for the management of resources. If viable business models for the combination of these two

technologies can be found, a true commodity grid can become reality. An essential principle of any viable

business model will be to secure the flexibility for service providers to manage and provision services. The

provider consumer relationship is encapsulated within a service level agreement (SLA). We propose that

this SLA contains terms that only relate to business level objectives (BLO). Deployment and management

details of a service are hidden by virtualisation in the provider’s domain and therefore should not be ex-

pressed in the SLA. The SLA will become key to build confidence in the business relationship between pro-

vider and consumer and a differentiating factor between providers in a market place.

1 INTRODUCTION

The modern enterprise software industry is driven by

a number of competing trends: cost reduction; faster

application introduction; and faster reaction to

changing markets. Monolithic applications are being

replaced with modular Service Oriented Architec-

tures (SOA) that can supply services on demand.

The provision and consumption of services will be-

come one of the key interactions between organisa-

tions. To enable rapid reactivity, resources and ser-

vices need to be managed and provisioned in a dy-

namic and automated fashion. The management in-

frastructure is further complicated when services

exist in different management domains.

To deal with this we rely on a management

framework in which the confidence of the consumer

is established through a contract with the provider of

the service. Such contracts, commonly known as

Service Level Agreements (SLA), set out the quality

of service (QoS) and the terms and conditions that a

consumer and provider of a service have agreed

(Mitchell & Mckee 2005). The SLA also specifies

how the service is priced and the compensation

terms if the SLA is violated. In a service oriented

computing landscape, every service needs to have a

SLA.

Traditionally the QoS terms of a SLA focus on

low-level technical attributes of the services. While

technical minded service consumers understand the

low-level aspects of a service, service consumers in

general cannot be expected to understand this level

of detail. They only need to know if the service

meets their business needs, how to interact with the

service and the terms the service is provided under.

All of these terms need to be agreed and are there-

fore included in the SLA. The SLA must be ex-

pressed in terms much closer to the business than the

technical requirements. At some point the terms of

the SLA need to be mapped into the technical con-

text of the service so that the service is provisioned

and managed accordingly.

In this paper we discuss B2B service provision-

ing and argue that the service consumer should not

have any visibility or control over the technical as-

pects of service provisioning and management. We

highlight the value of excluding the technical aspects

of a service from the SLA and why it is important to

let the service provider assume responsibility for the

mapping of the SLA into the technical context of

service provisioning.

123

Masche P., Mckee P. and Mitchell B. (2006).

THE INCREASING ROLE OF SERVICE LEVEL AGREEMENTS IN B2B SYSTEMS.

In Proceedings of WEBIST 2006 - Second International Conference on Web Information Systems and Technologies - Society, e-Business and

e-Government / e-Learning, pages 123-126

DOI: 10.5220/0001245101230126

 SciTePress

2 GRID SPECIFIC CHALLENGES

Although synergies between SOA and grids exist,

major barriers to the use of these complementary

technologies still remain. Perhaps one of the largest

problems is finding viable business models for the

use of resources and services in this new flexible

architecture. SLAs for services could be the key

mechanism to enable such viable business models

(Mitchell & Mckee 2005). While traditional grids

assume mutual cooperation between organisations, a

commercially viable grid needs to be governed by

SLAs to replace the assumption of cooperation. To

be of use for commercial grids, SLAs need to ad-

dress the needs of all stakeholders in the commercial

relationship.

Another challenge in the commercial environ-

ment in having to cross organisational boundaries

where the two parties may be competitors. This is a

problem new to the grid and is neglected in many

current grids. Competition influences what monitor-

ing and management access a service consumer will

get to a service. Hence, the structure of a SLA that

addresses the needs of inter-organisational grid ser-

vices will be different to that anticipated in the cur-

rent grid community.

The Provider and consumer of a service need a

mechanism to achieve a similar understanding of the

meaning of the QoS terms contained in the SLA.

Ontologies are increasingly proposed as a solution to

this problem. We believe that a single ontology that

covers all potential applications of SLAs will not

exist any time soon, if ever. Domain specific ontolo-

gies exist in a number of industrial sectors such as

banking and insurance. These ontologies are cur-

rently unlikely to converge and some of these are

proprietary.

The field of semantics and ontologies is an area

of great research activity. Whether there will be suit-

able ontologies for SLAs remains to be seen.

3 DIFFERENT VIEWS OF A

SERVICE

There are two key motivations for a business to out-

source service provision.

The benefit of using commodity services from

another party is a reduction in cost due to the pro-

vider being able to take advantage of the economies

of scale, and a reduction in the number of in house

support staff required.

In the case of specialist services organisations

benefit from access to services on an occasional ba-

sis that they could not afford to purchase under tradi-

tional business models. These new business models

may include software as service or “pay-as-you-go”

type pricing.

Both of these categories of services speed the

development of new applications as resources are

provisioned and deployed on demand by service

providers.

In order for a service provider to deliver these

benefits to the customer, he needs the flexibility to

manage his resources to make improvements to effi-

ciency and utilisation that give him a viable role in

the value chain.

Such flexibility for the service provider is only

achievable by reducing the technical detail that is

exposed to the customer and their role in service

management. Resource virtualisation creates a natu-

ral boundary between technical detail and service

functionality. After all, the service provider is

probably best placed to understand and manage the

technical aspects of a service and the customer un-

derstands best the business context in which the ser-

vice is being used.

When the operational aspects of the service are

hidden by virtualisation, the business context in

which a service operates will become the main

source for requirements of the service. Therefore we

propose that customers will look for services in

terms of the business level objectives (BLO) they

wish to fulfil.

Both the customer and the service provider have

their own BLOs. The service provider’s BLOs in-

clude operating the service profitably. The cus-

tomer’s BLOs were touched on earlier and may in-

clude cost reduction and responsiveness. Both the

customer’s and the provider’s BLOs are commer-

cially sensitive and are unlikely to be shared.

The common information that is shared between

the two parties will be in the SLA. The SLA will be

the binding contract that creates the business rela-

tionship. The BLOs of both service provider and

customer will directly influence the content of the

SLA.

Once the SLA has been agreed, it is the service

provider’s task to see that the service is provisioned

and managed according to the terms of the SLA.

The relationship and interactions between the

customers and provider are firmly placed in the busi-

ness space. The customer is not involved in these

technical aspects of the service as the SLA provides

the guarantees that should satisfy him.

WEBIST 2006 - SOCIETY, E-BUSINESS AND E-GOVERNMENT

124

In contrast to the customer, the service provider

needs to understand the technical aspects of the ser-

vice. In order to decide how to deploy provision and

manage the service, the provider needs to consider

two things: the terms of the SLA and his BLOs.

From this he can derive the policies that are used to

manage the service. These policies will form part of

a policy-based management framework that uses

event-condition-action type policies (Sacks et al

2003). Events originating from the service that need

to be passed to the customer must cross the border

between technical and business perspective at the

service provider. To ensure that the customer under-

stands the event, it must be placed into context. This

context is encapsulated within the SLA and therefore

events have to flow across the service provider’s

business space before they are passed on to the cus-

tomer.

As all management activity of the service is hid-

den from the customer how can the customer be con-

fident that the service provider is not violating the

SLA terms? To answer this it is important to under-

stand the SLA content.

A common error is too complex an SLA as dis-

cussed in (Twing 2005).

“Poorly structured SLAs can lead to interesting,

problematic and unintended results. One common

mistake is to create too many SLAs. This can dilute

the effect of the critical few drivers that most affect

the business.”

We therefore believe that the terms in a SLA

need to be set in a business level language describ-

ing performance level guarantees that are directly

linked to BLOs. Many of these will be perceivable

by the customer during service consumption, but

some may not. One way of solving this would mean

to give the customer access to technical detail. We

strongly believe that this should be avoided, as pro-

viders need to maintain the flexibility to provide

services dynamically and efficiently utilise their

infrastructure. Both of these are vital to ensure the

provider has a viable business model. We believe

that the customer needs to trust the service provider

to be willing to enter a business relationship. Our

proposed solution draws on existing industry prac-

tice of a rigorous auditing process.

By describing and offering services in business

language it is easier to compare service functional-

ity, especially for non-technical users. This compa-

rability enables customers to “shop around” for best

offers on similar services.

If the service consumer has the desire and tech-

nical understanding to set requirements against the

technical performance of the service, those require-

ments can be expressed in more technical terminol-

ogy. However, the provider may impose extra condi-

tions. As the provider has to give up some flexibil-

ity, it will cost more to provide the service. Secondly

the provider might not provide an open view to the

management information of his services but instead

may filter the events passed on to the customer to

maintain confidentiality. For example, the mere ex-

istence of an event may already disclose sensitive

information to competitors.

SLAs will be used to manage the risk and expec-

tations of both parties. They will become increas-

ingly important if a market is to develop.

4 CURRENT SLA

SPECIFICATIONS

Within the GRID research community a number of

efforts have been made to define the structure and

content of SLAs. The two leading efforts within the

web service community are WSLA (Ludwig 2003)

and WS-Agreement (Global Grid Forum 2004).

While each of the two incorporates some useful and

necessary features of a SLA, neither of them ex-

presses all that needs to be in a SLA.

Another interesting approach to create precise

SLAs is SLAng (Lamanna et al 2003 & Skene et al

2004). A SLA in SLAng describes the two involved

parties and the responsibilities of service consumer

and provider. SLAng is designed for a specific sce-

nario and contains fairly rich detail of the service

and how it is run.

We believe the focus of a SLA needs to be the

business objectives of the consumer.

The currently proposed SLA structures, WSLA,

WS-Agreement and SLAng, are all too focused on

the technical aspects of a service and do not attempt

to cover the service’s business aspects. We believe

SLAs should also contain non-functional terms.

These are important to build the business relation-

ship with the customers and provide a differentiating

factor between service providers.

The EU IST 6

Framework project NextGRID

contains work in the area of SLAs. The current focus

in NextGRID is on creating a representation of

SLAs that contains both functional and non-

functional terms.

THE INCREASING ROLE OF SERVICE LEVEL AGREEMENTS IN B2B SYSTEMS

125

5 CONCLUSIONS

The coming together of SOA and resource virtual-

isation can make the true commodity grid a reality.

A remaining obstacle to this goal is the creation of

viable business models for participants. SLAs could

become the key mechanism to enable viable busi-

ness models as they can replace the assumption of

mutual cooperation that inhibits commercial adop-

tion of grid.

Providing and managing services across organ-

isational boundaries imposes some new confidential-

ity problems that are largely ignored in current grids.

For both, customer and service provider to ex-

ploit the benefits of outsourcing, the service provider

needs to have flexibility in provisioning and man-

agement of his services.

We proposed to view the service from different

perspectives that distinguish between the customer’s

and the service provider’s view. The shared view is

defined by the SLA and will principally contain

business terms.

The service provider uses his business level ob-

jectives and the SLA to derive policies to provide

and manage the service. Information to the customer

about the service will flow in the form of events,

which are placed into a business context before they

are forwarded to the customer. This enables the cus-

tomer to easily assess the business impact of any

SLA violation, and thereby contributes to more real-

istic penalty agreements.

The structure and role of SLAs in a B2B system

must allow for virtualisation of the providers re-

sources. SLAs are only to be expressed in terms of

BLOs. The SLA mechanism helps to manage the

risks and expectations of both service provider and

customer.

SLAs can enable comparison between services

on business level and non-functional SLA terms and

enhance the ability of service providers to differenti-

ate their products.

Work under way within the NextGRID project is

attempting to produce a mechanism that provides

both a framework that builds confidence and allows

providers to manage offers within a market place. It

is hoped that this and other work in NextGRID will

lead to an architecture that can support the viable

business models for a commercial grid as identified

in this paper.

REFERENCES

Global Grid Forum (GGF) 2004, Web Service Agreement

Specification (WS-Agreement), viewed 11 November,

2005,

<http://www.gridforum.org/Meetings/GGF11/Docume

nts/draft-ggf-graap-agreement.pdf>.

Lamanna, D & Skene, J & Emmerich, W, 2003, SLAng: A

Language for Defining Service Level Agreements,

viewed 11 November, 2005,

<http://www.cs.ucl.ac.uk/staff/w.emmerich/publicatio

ns/FTDCS03/slang.pdf>, in Proceedings of the 9th

IEEE Workshop on Future Trends in Distributed

Computing Systems. pp. 100-106. IEEE Computer So-

ciety Press.

Ludwig, Heiko et al 2003, Web Service Level Agreement

(WSLA) Language Specification, IBM, USA.

Mitchell, Bryce & Mckee, Paul 2005, SLAs A Key Com-

mercial Tool, eChallenges e-2005, Ljubljana, Slove-

nia.

Sacks, L., Prnjat, O., Olukemi, T., Ching, A., Fisher, M.,

Mckee, P., Georgalas, N., Yoshii, H., Active Robust

Resource Management in Cluster Computing Using

Policies, in Journal of Network and Systems Manage-

ment, vol. 11, no. 3, September 2003.

Skene, J & Lamanna, D & Emmerich, W, 2004, Precise

Service Level Agreements, viewed 11 November 2005,

<http://www.cs.ucl.ac.uk/staff/W.Emmerich/publicatio

ns/ICSE2004/icse04.pdf>, 26th International Confer-

ence on Software Engineering (ICSE2004), Edin-

burgh, Scotland, UK, May 2004

Twing, D., Are you savvy about SLA negotiations?,

viewed 21 November 2005,

<http://www.networkworld.com/newsletters/asp/2005/

1003out1.html>, Network World.

WEBIST 2006 - SOCIETY, E-BUSINESS AND E-GOVERNMENT

126