Storage: Picking a bit bucket: SAN and NAS

Whether you are managing operations at a call-letter station with a half-dozen edit stations, or at a broadcast giant with hundreds of editors, architecting the storage supporting your facility’s file-based workflows is a business-critical task. Video editing places higher demands on storage than any other file-based application, requiring streaming performance for large files starting at 3.5MB/s for SD to 165MB/s per stream for uncompressed HD. And, with today’s higher-resolution formats, streaming video data demands even more performance from storage systems, with 4K requiring 1210MB/s per stream — 7.3X more throughput than HD.

Traditionally, this level of performance could only be met by high-performance disk storage directly attached to the editing workstation. The downfall of directly attached storage (DAS) is that it silos content on to individual computers. (See Figure 1) Sharing large media files between editors or moving the content to the next step in the workflow requires manually copying files across the network or resorting to the “sneakernet” solution of copying content on removable media to move it along the workflow. The result is expensive: Duplicate copies of large files double the storage capacity, and waiting for file transfers reduces the productivity of highly-paid editors.

High-performance shared storage systems were designed to solve this problem, and today there are two principal options: shared Storage Area Network (SAN) file systems and scale-out Network Attached Storage (NAS) file systems. (See Figures 2 and 3.) Based on fundamentally different architectures, each offers advantages and disadvantages that should be carefully considered before choosing a storage system for your workflow.

Block-based storage over Fibre Channel

With a SAN, a pool of high-performance storage is divided and allocated to individual servers. Users and applications can only access storage through allocated servers. This works well for databases, but not for media workflows where files are shared by teams working on different workstations.

Shared SAN file systems break the silos by adding file system functionality without adding a file system layer. (See Figure 4.) Access to data on the shared volumes is carefully controlled for data integrity, often by a separate server that manages file locking, space allocation and access authorization. Placing this server outside the data path, instead of between the client and the storage, eliminates a potential bottleneck and improves overall performance of the storage solution.

To deliver blocks over a network fast enough, SANs also use a storage-specific network standard, Fibre Channel, with its own dedicated switches, cables and protocol. The Fibre Channel protocol delivers SCSI commands between the server and the SAN’s disk systems just as it would a locally attached disk.