ROST: Remote and hot service deployment with trustworthiness in CROWN Grid

Abstract

The main goal of our key project, the CROWN Grid, is to empower in-depth integration of resources and the cooperation of researchers nationwide and worldwide. CROWN exploits a service-oriented architecture based on OGSA. In CROWN, remote service deployment is highly desirable. To the best of our knowledge, however, there is no successful solution to ensure the enabling remote and hot service deployment in grid systems. Traditionally, remote deployment is supported in a cold fashion, which results in many disadvantages, such as low efficiency. Moreover, since the deployer and the target container may be in different domains, great security challenges arise when a service is deployed to a remote container. In this paper, we present ROST, an original scheme of Remote and hOt Service deployment with Trustworthiness. By dynamically updating runtime environment configurations, ROST avoids restarting the runtime system during deployment. In addition, we include trust negotiation in ROST, which greatly increases the flexibility and security of the CROWN Grid. ROST has been successfully implemented. We conduct comprehensive experiments with real applications, and the results show that ROST is viable and significantly improves the service efficiency and quality of CROWN. We believe that the wide deployment of ROST would also benefit other grid systems.

Introduction

Grid computing promises to enable coordinated resource sharing and problem solving in dynamic, multi-institutional virtual organizations [17]. As grids are very heterogeneous environments, there exist various resources including networks, computers, storage and devices, and the like, to better tackle the heterogeneity problem. Service-oriented grid architecture has recently been introduced, which is widely considered to be the future of grid computing [19]. Built on web services, OGSA [18] is the de facto standard for building service grids, in which various resources are encapsulated as services with uniform user interfaces.

The main goal of our key project, CROWN (China R&D Environment Over Wide-area Network) Grid, is to empower the in-depth integration of resources and cooperation of researchers nationwide and worldwide. The CROWN project was launched in late 2003. As illustrated in Fig. 1, a number of universities and institutes, such as Tsinghua University, Peking University, Chinese Academy of Sciences, and Beihang University, have joined CROWN, each contributing 50–100 computing nodes. More universities and institutes are invited to join the CROWN Grid. CROWN will also be connected to some world-famous grids, such as GLORIAD [7] and PRAGMA [8].

In past years, many key issues in grid computing have been extensively studied, such as information services, resource scheduling, and security issues. As an important issue, remote and hot service deployment, however, has not been fully addressed. Before a service is ready for invocation, it must be deployed in a service container, which provides a runtime environment. A grid is a highly distributed environment, in which numerous domains could be involved. The domains are usually geographically dispersed. It is highly desirable for a contributor to deploy its services within remote service containers for multiple purposes. For example, in the CROWN Grid for applications in bioinformatics, there are many computing intensive applications such as BLAST. A computing node could easily become overloaded when multiple jobs arrive in a short time period. The heavy load can be balanced if the node is able to deploy one or more BLAST service replicas to remote nodes and then redirect some jobs. Similar requirements also exist in many other grid applications.

Traditionally, remote service deployment is supported in a cold fashion, which means that to deploy a new service, the runtime environment needs to be restarted. This results in many disadvantages because previously running services must be stopped, and they may have to resume or even restart their jobs, causing significant overheads. Therefore, hot service deployment, which does not need to restart the runtime environment while deploying services, has become increasingly important. With the availability of remote and hot service deployment, many applications will benefit, such as load balancing, job migration and so on.

Service deployment is actually not a new issue. Similar demands also exist in mobile agents [11] and active networks [13]. To the best of our knowledge, however, there is no successful solution to enable remote and hot service deployment in grid systems. The most updated Globus Toolkit (version 4) [4], the de facto standard for grid middleware, does not yet provide the function of remote and hot service deployment. This may be due to the great security challenges arising when a user deploys a service to a remote container. Here, we call a node a deployer, which intends to deploy a service, and the remote service runtime environment the target container, which is responsible for running and managing services being deployed. Without proper security mechanisms, a service provided by a deployer may be malicious, and the target container may be rogue or fragile. Also, the security policies of the deployer and the container could be incompatible. In an open grid environment, we cannot expect any deployer and its corresponding target container to set up a required trust relationship in advance. Moreover, it is too costly to always have to build trust across domains based on the traditional PKI infrastructure during remote deployment.

In this paper, we present our original work, ROST (Remote and hOt Service deployment with Trustworthiness), which achieves its goal by dynamically updating the runtime environment configurations. ROST avoids restarting runtime systems during remote deployment. Moreover, we include trust negotiation in the ROST scheme, which greatly increases the flexibility and security of CROWN. The major contributions of this work are as follows:

• We identify the necessity of remote and hot service deployment in service grids, and their challenges.

• We propose an effective approach, ROST, to enable remote and hot service deployment. Also, we add trust negotiation into the scheme to meet general security requirements for grid environments.

• We implement the ROST scheme in the CROWN Grid with real applications, and evaluate the performance of ROST by means of comprehensive experiments.

The rest of this paper is organized as follows. In Section 2, we present our proposed approach to ROST in detail. The implementation experiences are introduced in Section 3. We present our experimental methodology and performance evaluation of ROST in Section 4. We discuss related work in Section 5 and conclude this work in Section 6.