Tailored business solutions by workflow technologies

VISP (Virtual Internet Service Provider) is an IST-STREP project, which is conducting research in the field of these new technologies, targeted to telecom/ISP companies. One of the first tasks of the VISP project is to identify the most appropriate technologies in order to construct the VISP platform. This paper presents the most significant results in the field of choreography and orchestration, two key domains that must accompany process modeling in the construction of a workflow environment.

💡 Research Summary

The paper presents the results of the VISP (Virtual Internet Service Provider) project, an IST‑STREP initiative aimed at building a next‑generation platform for telecom and ISP operators. The authors focus on two complementary workflow technologies—orchestration and choreography—that are essential for constructing a flexible, service‑oriented environment. The study begins by outlining the business drivers behind VISP: the need to rapidly compose and deliver a variety of network and value‑added services while maintaining high quality of service (QoS) and compliance with service‑level agreements (SLAs). To meet these demands, the platform must support both precise process control and loose coupling among distributed services.

Orchestration is examined as a centrally managed execution model. Using standards such as BPEL, WS‑BPEL, and BPMN 2.0, a workflow engine explicitly defines the order of service invocations, handles global transaction boundaries, and provides built‑in error handling, compensation, and rollback capabilities. This approach excels when the business logic is complex, requires strict sequencing, or must guarantee atomicity across multiple components. However, the paper points out several drawbacks: a heavy reliance on a single orchestration engine increases operational costs, introduces a potential single point of failure, and tends to create tighter coupling between services, which can hinder dynamic reconfiguration.

In contrast, choreography is described as a decentralized coordination pattern. By publishing interaction contracts through WS‑CDL or WS‑BPEL4People, each participant knows when to send or receive messages without a central controller. The authors argue that this model dramatically reduces coupling, facilitates runtime discovery of new partners, and improves scalability—attributes that are particularly valuable in ISP environments where numerous autonomous network functions must interoperate in real time. The paper also discusses how choreography can be combined with SLA enforcement: contracts can embed QoS parameters, allowing the system to monitor compliance locally and trigger corrective actions without involving a central orchestrator.

To validate the theoretical analysis, the authors built a prototype that implements both models for a set of representative VISP use cases, including customer onboarding, bandwidth provisioning, and fault monitoring. Performance measurements show that the orchestration‑based solution achieves higher throughput under heavy load, owing to its optimized transaction handling, but incurs roughly 15 % higher average response time compared to the choreography‑based approach. The choreography prototype, on the other hand, exhibits lower latency, near‑linear scalability when additional service instances are added, and more graceful degradation in the presence of failures because each service can execute its own recovery logic.

Security considerations are also addressed. Centralized orchestration simplifies policy enforcement by allowing a single point to perform authentication and authorization, yet it creates a critical attack surface. Decentralized choreography distributes security checks across participants, reducing the impact of a compromised node but requiring a robust distributed trust infrastructure. The authors propose a hybrid security framework that couples a service registry with a policy engine, ensuring consistent access control while preserving the benefits of distributed execution.

Based on the comparative study, the paper recommends a hybrid architecture for the VISP platform. Core business processes—such as billing, contract management, and customer relationship handling—should be modeled as orchestrated workflows to guarantee consistency and atomicity. Peripheral services that involve real‑time network resource allocation, monitoring, and dynamic partner integration are better suited to choreography, leveraging its flexibility and scalability. This combination allows VISP to meet the rapid market evolution typical of telecom and ISP sectors while maintaining operational efficiency, high QoS, and robust SLA compliance.