We have implemented a virtual machine (VM) for Java which executes on a cluster. Our cluster VM completely hides the cluster from the application, presenting a single system image (SSI) (i. e., the application sees a traditional virtual machine). At the same time it leverages the cluster to achieve improved performance for a range of applications. We show how the flexibility and constraints of the Java Virtual Machine (JVM) Specification [7] impacted the design of our cluster VM. We describe issues related to class loading and distribution-aware implementations of the bytecodes. We also point out the limits on providing a solution for completely transparent distribution of multi-threaded Java applications if one does not modify the VM or the core classes. © Springer-Verlag Berlin Heidelberg 2001.
CITATION STYLE
Aridor, Y., Factor, M., & Teperman, A. (2001). Implementing java on clusters. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2150, 722–732. https://doi.org/10.1007/3-540-44681-8_104
Mendeley helps you to discover research relevant for your work.