Several companies have claimed to have software-defined storage solutions; however, these solutions were not truly software-defined. Hammerspace founder David Flynn admits that he is solving a problem that he is partially responsible for creating. David was one of the two founders of a company the eventually became Fusion-io; SanDisk acquired Fusion-io in July 2014.
Serving and managing data across traditional workloads, microservices, on-premises, or any number of clouds has always been a challenge. Data Gravity attracts services, applications, and people around large volumes of data around storage silos. As data continues to build mass, services and applications are more likely to be drawn to the data, rather than vice versa. Placing large volumes of data in multiple locations around the globe to be closer to the applications that need to access it, for workloads like real-time analytics, is both impractical and expensive.
Data should be organized the way data is used, not the way data is stored. With Hammerspace the metadata control plane is separated from the data plane so that data can be abstracted from the infrastructure, similar to the way SDN creates an overlay and underlay layer. This allows the metadata to be placed where it needs to be, close to the applications that need access to data, essentially delivering data-as-a-service. This is the first time I have seen an actual Software-Defined Data Management Solution that could serve data anywhere.
Applications access data through Hammerspace, not the storage directly, which allows the policies to be applied globally regardless of the infrastructure that data happens to live on. These policies allow role-based access to the data, and protection for the data including the ability for global snapshots, global undelete, and disaster recovery. A policy can be set so that much like the recycle bin in windows, deleted enterprise data can be self-recovered by users for a period of time.
Because Hammerspace is only replicating metadata across all sites workloads can be deployed with tremendous agility, without disruption like downtime or app reconfiguration. What Kubernetes does for orchestrating workloads by declaring intent and letting the system figure out how to deliver on the request, Hammerspace does for data at file-level granularity – orchestrating and automating a hybrid cloud file data layer that spans storage, sites, and clouds.