ARC and gLite Interoperability in ATLAS Sites

In the Worldwide Large Hadron Collider Computing Grid several sites deploy both gLite and ARC middlewares. In this manner they are simultaneously providing resources to more than one federation of sites. In this article the interoperability between the two middleware flavors at sites that provide distributed resources to the ATLAS experiment is described. Some motivations for dual deployments, the coexistence of a gLite computing element and an ARC front-end on the same local resource manager system, issues related to an ARC front-end with a local storage element, consistent monitoring, accounting and ticket handling are discussed. Finally needs for increased interoperability, based on experience of ATLAS sites, are given.