Ceph-Based Storage Holds Efficiency Promise

IT professionals who must support varied types of workloads within the same cluster probably have already discovered Ceph, an open source software-defined storage technology. Experts routinely praise Ceph for its ability to manage huge amounts of data in one unified storage cluster composed of object, block and file storage.

Fujitsu knows this as well. It has introduced a technology it calls “Dataffinic Computing” that aims to reduce bottlenecks that can occur when storage systems become overwhelmed and servers can’t keep up. The technology, developed using Ceph, is designed to support high-speed processing of massive data in distributed storage. The new technology distributes and collects data by connecting multiple servers through a network. This allows storage systems to process large volumes of data at high speeds, creating greater levels of efficiency. It also can handle unstructured data such as log data and video.

“For Ceph-based storage systems, this is an example of how underlying server compute resources can be leveraged to support a form of convergence, not to mention delivering on some of the previous software-defined storage premises,” said Greg Schulz, president of StorageIO.

One of the most important features of the new technology is the way it handles unstructured data. It breaks down unstructured data into “natural breaks,” and stores the data in a way in which the pieces can still be processed individually. Metadata stays attached to each piece of data.

Another important feature is adaptive resource control with storage functionality and data processing capability. The function controls data processing resources and how they are allocated in a way that doesn’t slow down storage performance.

“In theory, the concept should be a hybrid home run for boosting data intensive analytics and other rich content applications,” Schulz said.

While other vendors have previously deployed solutions similar in concept, this is the first for Ceph-based systems. Watch this space, Schulz said. Expect the evolution to continue.