This column defines the fourth of five storage networking architectures, the "in-band telecom" model. This model represents an emerging segment of storage networking environments today and possesses three distinct characteristics. First, it manages where data lies on the storage devices. Second, it does this management within the data path of the storage network itself. Third, the switch itself can perform advanced functions such as routing and quality of service.
Discussed in an earlier column in this series, "in-band" technology can enable the discovery and assignment of LUNs on any storage array it manages. This methodology provides a central console to report on and manage data within the storage network itself. Further, it delivers this functionality without requiring the deployment of server agents or the use of APIs.
Also previously discussed, the term "telecom" is borrowed from the telephone switching model. Telecom switches contain more intelligence than their ESCON counterparts because they enable advanced features such as Quality of Service (QoS), security and trunking. The QoS functionality allows switches to prioritize traffic on the network, based upon information contained in the data packets themselves. The security function authenticates servers logging into the network while trunking enables multiple physical paths between switches to function as one logical path.
The "in-band telecom" model makes sense for organizations that have larger SANs they wish to link together while simplifying storage management, a problem created by the out-of-band telecom model. It takes much of the current pain of LUN masking and zoning out of the equation while creating an important new layer in the network from which to manage storage. This simplification of the management of storage helps explain some of the movement in the industry to implement this solution.
This "in-band telecom" model merits serious consideration for enterprises looking to deploy heterogeneous SANs using multiple vendors OSes and storage arrays. Yet one should approach this model with caution -- while it may free you from locking into either a host based or storage array vendor to manage storage -- it does lock you into the switch vendor who implements this technology.
Vendors involved with the development of these "in-band" switches are well aware of the potential of locking into their switches and, in the next column, we will look at a derivative of these existing models that helps to break the potential of vendor lock in at the switch level.
Read Part I: Design concepts for storage networks
Do you want to see more articles by Jerome Wendt or insights from noted industry observers? Visit the complete Bits & Bytes column library.
About the author
Jerome Wendt is a independent writer and storage analyst specializing in Open Systems storage and storage area network technologies. He currently manages storage and explores new storage technologies for a large organization in this capacity. Jerome may be reached at