Release 1.5.0 of the Vitastor software storage system with support for the cluster file system (VitastorFS) has been published. Vitastor is a distributed block software data storage system, that is, a storage of images of virtual machines or container disks, developed by the author since 2019. As of release 1.5.0, it is also a clustered file system. VitastorFS is mounted using the NFS 3.0 protocol from a locally or remotely launched server/servers.
Vitastor is aimed primarily at SSD and SSD+HDD clusters (but also works in HDD-only configurations) with at least a 10 Gbps network, supports TCP and RDMA, and on good hardware can achieve 4KB read and write latency on ~0.1 ms, which is about 10 times faster than most popular SDS and internal SDS of large cloud providers - both Ceph and typical cloud SDS provide write latency of 1-2 ms even on very fast drives.
For example, in a recent test with Samsung PM9A3 NVMe drives and Mellanox ConnectX-4 RDMA NICs, the ideal write latency was >0.1 ms when using replication and 0.1 ms read / 0.2 ms write when using error correction codes.
Vitastor supports the QEMU driver, NBD and NFS protocols, Proxmox, Kubernetes CSI, OpenStack drivers. An experimental Kubernetes operator and Prometheus exporter are also available for Vitastor.
Vitastor server components are distributed under its own βnetwork copyleftβ license VNPL (Vitastor Network Public License). The idea of ββVNPL is to extend the effect of copyleft not only to modules that are explicitly associated with the code, but also to services interacting with Vitastor over the network, specially created for it (for example, any bindings for internal use in a commercial cloud). The rest of the VNPL text is copied from the GNU GPL 3.0.
Source: opennet.ru