The BEAR Blog

With node-oriented clustered block storage systems, the data redundancy overheads can be significant, and contingent on mirrored data. Some architectures have managed to reduce these overheads, but not as substantially as, for example, RAID 5. But “stacked controller” products can be configured with data mirroring as well as RAID 5, and so they do not share the same disadvantage. Clustered storage can also offer a paler version of response-time performance, in contrast to the throughput performance. Substantial input- output overheads are generated for reading and writing even small blocks of data, which can bring about lengthy response times to particular requests. Clustered block storage is not ideal for transaction-oriented, small block, randomly accessed workloads, although “stacked controllers” are again the exception and can boast very good response times even for these workloads.
READ MORE

In a game-changing shift in the data storage field, a move towards clustered storage architecture is rewriting the conventions of storing and accessing data. Vendors are now looking increasingly towards these clustered technologies to solve four key problems that beset storage managers. Clustered systems, in fact, are designed with the exclusive objective of addressing these four problems.
READ MORE