Networking, storage and multiplatform production
Oct 5, 2007 3:00 PM
To the desktop
Another point to consider is that the convenience of centralized storage is offset by the higher network bandwidth required to meet the needs of multiple users. For example, centralized asset management use proxies and thumbnail images that can be viewed from anywhere within an organization. This implies that they will be transferred or streamed over the corporate network.
From a network bandwidth capacity perspective, allowing for collisions and assuming 50 percent bandwidth use, layer 3 GigE distribution level routing may support 500 1Mb/s streams concurrently, at least theoretically.
But keep in mind that all other network traffic, such as e-mail, the Internet, office applications and content file transfers, may be traveling on the network at the same time. As collisions occur, the network gets continually slower. Under these circumstances, GigE is not very fast, causing content delivery to sputter.
A network design may work fine under normal working conditions, but a standard process in commissioning a system is to stress the network until it fails. It’s better to do this before finalizing the network design than the infrastructure crashing when it is needed most.
Characteristics of a routable network system that must be considered when choosing a routing protocol include:
- Optimality: Metrics used by a routing protocol must be relevant and produce acceptable file transfer performance.
- Simplicity: Router overhead (adding additional bytes to a payload) and efficient use of routing resources helps maintain a stable and reliable network.
- Robustness: The routing protocol must meet the file transfer requirements. A flat network, hierarchical network or combination of the two may be best.
- Convergence: The time it takes to update routing tables and calculate new link metrics impacts meeting availability and service level requirements.
- Flexibility: Algorithms used by the routing protocol must adapt to the changing dynamics of the network.
Don’t be satisfied if a network design initially appears to be performing acceptably. The capabilities of the design will not be tested until more storage, routing devices and workstations are added to the network.
A paradigm shift
Because storage and distribution are interdependent in a media network, the two are being merged into one structure. High performance storage is incorporating ASIC-based hardware routing functionality. Another methodology replaces Fibre Channel interconnections with storage blades that use rack backplane interconnection. Some vendors offer total solutions that support SDI and AES3 I/O connected to scaleable storage systems, yet these systems are designed for production that supports single-channel distribution.
Researchers and developers continue to explore methodologies to improve storage network systems. New technologies such as holographic storage, interface techniques like Inifiniband and network technologies like 10GigE are decreasing access time and file transfer latency in all media network topologies. One day, quantum computing will be the norm.
Perhaps what is needed is not just a better implementation of the same physical technologies, but a paradigm shift that will adapt to the new technologies on the commercialization-commoditization horizon.
People will continue to communicate and receive information through an increasing array of technologies. Soon, broadcasters will have more platforms than ever to maintain. Production networks installed today will have to be designed with enough agility to adapt to the needs of parallel production for multichannel delivery now and in the future.
| Want to use this article? Click here for options! |  |
