Dynamic Circuits for Global Research: Reserve your Space on the AutoBAHN
By Afrodite Sevasti, GRNET
Modern researchers grow evermore demanding in their use of research and education network infrastructures.
Collaborative work means they increasingly need to transport large volumes of scientific data at fast speeds and with guaranteed levels of service. The GÉANT2 “Automated Bandwidth Allocation across Heterogeneous Networks”—or AutoBAHN— is a system designed with just that in mind.
AutoBAHN ensures that demanding network users and next generation applications can get guaranteed access to service of a guaranteed quality, overcoming the challenges presented by today’s physically and technically disjointed networks. Still in its pilot stage, AutoBAHN is already delivering results: users can access a fully operational circuit of guaranteed quality within minutes of submitting their request (subject to resource availability).
The case for dynamic circuit services
Although Internet Protocol (IP) networks provide “always-on” data transfer services, they face many ongoing challenges, especially when it comes to ensuring quality of service. IP networks simply cannot guarantee that adequate resources will be available to scientists who must transfer large amounts of data within specific time constraints.
This is a regular problem for researchers who are data streaming from different locations to a cluster for real-time correlation.
These limitations are inherent: the networks are heterogeneous, and their open nature means a large user base must compete for access. The alternative—to provide fixed circuits that exclusively service demanding research applications—is costly and often results in underutilised e-Infrastructures.
A dynamic circuit service is one way of overcoming these limitations in certain use cases. Such services identify and reserve resources over existing infrastructures, thus guaranteeing quantity and quality for a limited, yet tailored, set of conditions. When reserved resources are no longer required, they are released and become available to the next user.
Dynamic success
Automation is crucial to the success of a dynamic circuit service: middleware must manage much of the process, automatically reserving resources and assigning service levels. Availability and functionality are equally essential: the service must be end-to-end and cover multiple network domains. Users must be able to reserve resources in advance, and monitor their work as it progresses. This level of service requires coordination across domains, a service provided by the AutoBAHN system.
AutoBAHN complements, but does not necessarily replace existing network signaling and provisioning capabilities. Instead, it integrates new services for coordinated inter-domain provisioning, including AAI, inter-domain routing functionality, interdomain monitoring, and so on.
The system is simple: AutoBAHN interprets user requests and translates them into network requests. The networks automatically update AutoBAHN as to what services are available, their quality, and their policies for dynamic circuit delivery. Auto-BAHN instances at each network then use Web Services to accept dynamic circuits requests based on their currently supported topology.
Astronomically useful
AutoBAHN has already attracted a number of pilot use cases. A recent joint demonstration with the SCARIe project used AutoBAHN-enabled dynamic circuits to send pre-recorded data from radio telescopes across Europe to the DAS-3 cluster at the University of Amsterdam for correlation. This demonstration used the e-VLBI network.
This application serves to demonstrate a number of AutoBAHN’s salient points:
• dynamic circuits provide an effective way of working with the e-VLBI network, which does not operate continuously
• the locations of the telescopes vary, thus there is a need for data transfer from a wide variety of locations
• in an advanced software-based correlator setup, the software can be run on many distributed clusters, so dynamic circuits are needed to interconnect those clusters on a case-by-case basis.
AutoBAHN was developed as a pilot within the GÉANT2 project, which is co-funded by the European Commission as part of the Sixth R&D Framework Programme (FP6).
The SCARIe project is a collaboration between the JIVE (the Joint Institute for Very Long Baseline Interferometry (VLBI) in Europe), the University of Amsterdam and SARA, and is funded by the Netherlands Organization for Scientific Research. SCARIe aims to develop a distributed software correlator for real-time e-VLBI in conjunction with
advanced networking technologies.
