Single Instance Store

The performance of a solution depends on how well data is organized on disk, how fast data can be located to recreate previously seen content and by how large the usable disk is.

The Challenge

The challenge with de-duplication in WAN optimization solutions is that the databases that store the content in each peer device need to be synchronized. The simple approach is to partition the disks based on the number of connected peers and ensure the databases for the pair are in sync, this is known as a per peer data store. However this poses a scalability problem for the data center in a hub and spoke deployment. For example, if an email was sent to the entire company multiple copies would be stored on the data center appliance, one copy for each peer. This is very inefficient use of the acceleration history disk. With a per peer data store the options are clear:

1. Divide the data center’s disk by the number of peers. This makes the effective usable disk very small and therefore performance suffers.
2. Deploy multiple appliances in the data center such that each peer has reasonable store size. This is usually too costly.

Single Instance Store

The Certeon aCelera uses a Single Instance Store. This technology is built into the architecture and it keeps only a single copy of any content irrespective of the peer is being sent to, which protocol is being used and in which direction the data is being transferred. The benefit of this technology is that it allows the solution to scale to support the needs of a large deployment while delivering the performance that is required.

The Single Instance Store technology is critical when supporting an environment with users using a software accelerator for the individual user, otherwise the effective usable data store at the data center would be too small to be effective. The Single Instance Store technology also allows for complex deployment environments, such as meshed networks, to perform.