APPLICATIONS OF TECHNOLOGY:
- Cloud scheduling and Quality-of-Service management, workflow management and co-scheduling, storage architectures in cloud
- Large-scale data movement (giga- to terabyte) with on-demand bandwidth allocation
- Collaborative research across multiple institutions (e.g., accelerator-based physics, global climate studies, distributed applications with giga- to terabyte data communication needs)
- Network provisioning and advance resource allocation for predictable performance
ADVANTAGES:
- Provides users with reservation choices to meet their time and resource requirements
- Gives users flexibility to allocate and provision the network in advance
- Provides efficient methodologies to manage time-dependent network topology
- Compatible with current network reservations system
- Optimized for fast performance
ABSTRACT:
Scientists at Berkeley Lab have developed a flexible reservation algorithm for advance network provisioning. The algorithm finds communication paths in time-dependent networks with bottleneck constraints, factoring in parameters that the user provides. The system then offers reservation options that meet these requirements. For example, it may offer an early start time with relatively long transit time or a late start time with a shorter transit time. The intellectual property also includes a software library that can be used to integrate the algorithm into existing network reservation frameworks.
The Berkeley Lab scientists analyzed time-dependent networks with bottleneck constraints. The general solution to such analysis requires exponential time complexity. However, they developed novel approaches to reduce the amount of information that the algorithm must handle, thereby improving its efficiency. As a result, the algorithm is of polynomial-time complexity in the worst case scenario, and it is bounded by the number of nodes (i.e., routers) and the number of existing reservations in the desired time interval. Because of this relative simplicity, the algorithm, when tested, could produce a list of reservation options in less than a second for a network with 1000 nodes and several hundreds of reservations already made for the desired time interval.
Analyzing and moving massive datasets has become increasingly important in scientific and business applications. Existing network reservation systems establish guaranteed bandwidth of secure virtual circuits at a certain time, for a specific bandwidth and length of time. If the requested reservation cannot be granted, no further suggestion is returned back to the user, who must then use trial-and-error to find a reservation. As a result, these systems inefficiently consume users’ time and system resources and do not allow users to make optimal choices. The Berkeley Lab algorithm and software library overcomes these limitations.
DEVELOPMENT STAGE: Tested in network simulation; implementation in progress for Energy Sciences Network (ESnet).
STATUS: Available for collaborative research.
FOR MORE INFORMATION:
SEE THESE OTHER BERKELEY LAB TECHNOLOGIES IN THIS FIELD:
FastBit: Ultrafast Database Searching Using a Patented Indexing Method, IB-1852
Relationship Network Search Engine, JIB-2042
REFERENCE NUMBER: IB-2962
