CADAC
About
The CADAC will host a large collection of computational data residing on permanent disk space while being managed by the Storage Resource Broker (SRB) software at the San Diego Supercomputer Center (SDSC). Currently, we have 100TB of dedicated disk space, and could possibly expand the data collection to 300TB in the future.
The CADAC also offers access to powerful data-analysis hardware, composed of two large shared memory IBM nodes that are part of SDSC's Datastar. Both nodes are IBM p690s with 32 POWER4+ processors each, and 256GB of shared memory. Data analysis and graphic visualization software such as IDL and VAPOR are available on these nodes.
Guidelines
The goal of the CADAC is to provide a mechanism for publishing, sharing and analyzing large computational datasets. The available resources are intended for data-analysis of data in the CADAC collection. Thus, they should not be used to run simulations, but only to study their results.
To aid this, we encourage users to submit new simulations to the shared CADAC collection, in order to take advantage of our resources. The permanent disk space, automatic archival system, powerful data-analysis hardware, and ability to share data and tools with their immediate collaborators are incentives for users to publish their simulation data to the CADAC collection as early as possible.
SRB
The shared data in the CADAC is managed by the the Storage Resource Broker (SRB) software developed at SDSC. It provides a rich set of features including a single logical namespace, clients for several platforms, and access to multiple physical storage devices. For the CADAC, we have implemented a powerful system combining both file storage on disk, and automatic replication to tape in the High Performance Storage System (HPSS). Files in the shared data collection are readable directly from disk, while new and modified files can be automatically replicated to tape.
