Distributed and on-demand cache for CMS experiment at LHC

摘要

In the CMS [1] computing model the experiment owns dedicated resources around the world that, for the most part, are located in computing centers with a well defined Tier hierarchy. The geo-distributed storage is then controlled centrally by the CMS Computing Operations. In this architecture data are distributed and replicated across the centers following a pre-placement model, mostly human controlled. Analysis jobs are then mostly executed on computing resources close to the data location. This of course allow to avoid CPU wasting due to I/O latency, although it does not allow to optimize the available job slots. The continuous increase of the storage requirements (a factor 20 is foreseen for HL-LHC era), as well as the proliferation of models with not-owned and on-demand resources (on private or public cloud and HPC facilities) are pushing in favor of loosening up of the CMS computing model, looking for solutions that optimize both the amount of space centrally managed by computing sites and the CPU efficiency for jobs that run on storage-less resources. As matter of fact the Tier 2 storages, for the most part, could be operated as an unmanaged cache space, thus reducing by a large fraction the operational cost with an increase in flexibility. In this scenario, the cache space will look like a distributed and shared file system populated with the most requested data; in case of missing information data access will fallback to the remote access. We think that such caches should be clustered by geographical region, in order to leverage the national high bandwidth network from NRENs and minimize the amount of replicated data between sites. A geographically distributed caching layer will be functional also to a data-lake based model where many satellite computing centers might appear and disappear dynamically. It seems quite reasonable that a protection layer against a central managed storage might be a key factor along with the control on data access latency. In this