A service platform for development, deployment and runtime management of Real-Time Online Services
Contact person
Dominik Meiländer, d.meil@uni-muenster.de, University of Muenster, Germany
Description
The Real-Time Framework (RTF) is a middleware technology for high-level development of scalable Real-Time Online Services along the S-Cube Lifecycle Model. RTF supports three different parallelization and distribution techniques to scale Real-Time Online Services on multiple Cloud servers: zoning, replication and instancing. RTF was used in several industrial applications, e.g.: a multi-server port of the commercial action game Quake 3; a 3D online game Hunter developed by the game company Darkworks; a remote e-learning framework edutain@grid Virtual Classroom developed by the environmental consulting company BMT Cordah Ltd.
RTF is complemented by the Hoster Management Interface (HMI) that supports the transparent resource management for a running application, in particular the creation, controlling and monitoring of Real-Time Online Services. HMI offers a application providers a management interface for Real-Time Online Services which are implemented on top of RTF. Application providers can deploy their Real-Time Online Services on their own resources or in the Cloud.
For details of RTF and HMI, see: www.real-time-framework.com
Technical Information
The Real-Time Framework (RTF) provides to application developers a cross-platform C++ library with the following functionalities:
- An automated serialization mechanism, which liberates the developer from the details of network programming.
- A highly efficient communication protocol implementation over TCP/UDP optimized with respect to the low-latency and low-overhead requirements of ROIA. This implementation is able to transparently redirect communication endpoints to a new resource, if, e.g., parts of the ROIA are relocated to a new Grid/Cloud resource for load-balancing reasons.
- A single API for using different parallelization approaches: zoning, instancing, replication and their combinations, for a scalable multi-server implementation of ROIA.
- A fully automated distribution management, synchronization and parallelization of the ROIA update processing.
- A transparent monitoring of common ROIA metrics that is used by the management and business layer of the edutain@grid system.
Demo
A video demonstration is available at www.real-time-framework.com
Publications
RTF Tutorial for Online Game Development
Area
Software Engineering Life-Cycle, Service adaptation, Grid & Cloud Computing
Maturity Level
Prototype (for different scenarios)
Relationship with Future Internet and Internet of Services
Future Internet and Internet of Services with their Infrastructure-as-a-Service (IaaS) approach offer new opportunities for the execution of Real-Time Online Services and promise a potentially unlimited scalability by distributing application processing on an arbitrary number of resources given suitable adaptation mechanisms. IaaS allows for adding/removing resources on demand. This opens for Real-Time Online Services an opportunity to serve very high numbers of users and still comply with QoS demands. Despite a variable number of users, IaaS resources can be used efficiently if the application supports adding/removing resources during runtime. Hence, using Cloud Computing for resource provision and the Lifecycle model for implementing adaptable Real-Time Online Services complement each other.
Relationship with Cloud
Our middleware platform supports a high-level development approach for Real-Time Online Services and automatically distributes service processing on multiple Cloud resources. We have implemented a resource management system that uses Cloud resources for cost-efficient and dynamic up- and down-scaling of application sessions using dynamic load balancing based on application-specific monitoring values