Event Details

Deployment factors of orchestrated converged compute with Ceph block storage explored in this talk. The hyper-convergence server is a software-centric architecture that tightly integrates compute, storage, and networking on commodity hardware.  Many compute servers in the in data centers have internal storage disks, but block storage is provided by dedicated SANs. As a result, much of the local storage on compute nodes remains underutilized.  Ceph stores data on distributed commodity servers and provides interfaces for object and block storage. While Ceph is often run on dedicated servers, it can make more sense to converge OpenStack compute and Ceph on the same nodes and employ the unused local hard disks as drives for Ceph OSDs.  Converged Ceph can significantly reduce storage costs in comparison to proprietary iSCSI arrays. It allows for a scalable storage infrastructure that grows as compute capacity increases, and eliminates the need for dedicated Block storage system.