Broadcasters spend a lot of time researching storage these days — and with good reason. The conversion to digital and the ever-increasing number of computer video devices in facilities has prompted many broadcasters to design and operate IT-based facilities. Video is a large data type, so storing even modest amounts of it requires a lot of disk space.
There are three main storage solutions: direct attached storage (DAS), storage area networks (SANs) and network attached storage (NAS). (See “Comparing storage” on page 30.)
Direct attached storage
DAS consists of storage directly attached to a server or workstation, perhaps by a SCSI cable or Serial ATA cable. In the simplest case, files stored on one workstation may be shared with other computers on the network using file sharing capabilities built into the workstation OS. (See Figure 1.) The workstation user simply grants others permission to view files stored on the main computer.
DAS is inexpensive to implement. Today's modern computers provide all the necessary components to set up file sharing without purchasing any special hardware or software.
The complexity of DAS file sharing is low. System configuration is minimal, and once it is up and running, very little, if any, maintenance is required. DAS performance, however, can be poor to the point of being unusable, especially if video is involved. Performance can worsen if the workstation user is engaged in a complex task while an external user is trying to access a video file.
Flexibility with a DAS solution is limited. Typically, file sharing is only supported under one OS. File sharing between multiple operating systems causes the DAS configuration to be more complicated.
DAS solutions should never be used for a WAN or remote storage over the Internet. DAS file sharing is not as secure as other solutions, so it is imperative that it only be used with a LAN with a firewall between the LAN and the Internet to prevent local files from becoming visible to others outside the facility.
DAS file sharing is not particularly reliable. If a user needs a file on a workstation and the drive in that workstation fails, then the data will be lost unless steps are taken to mitigate this loss, through nightly backups, for example.
The loss of a drive will probably affect operations, and in most cases, it can take time before the data is restored. Configuring per-user access permissions for DAS can be challenging, especially if more than one computer is sharing files on a network.
Finally, as the name implies, DAS shares files. This means that a remote user accesses the entire file rather than only accessing the portion of the file being worked on.
Broadcasters looking for an inexpensive way to share small amounts of data over a small network should consider DAS, although they should take reliability into account.
Storage area networks
SANs have their roots in DAS. Initially, disks could not be mounted more than a few feet away from the computer they served. As storage requirements grew, it became necessary to externally mount disk arrays. By the time the cables (generally SCSI) were connected to all of the drives, there was little, if any, cable left over to run from the disk chassis to the main computer.
The first SANs allowed computer designers to break this constraint by implementing the SCSI protocol over another transport technology, typically Fibre Channel. As a result, drives could be mounted further away from the computer.
Today, SANs provide high-performance, remote-shared storage to multiple users. It does this, in part, by employing a dedicated network to move data. The SAN is not the same network used by client computers to access the Internet or to perform other tasks. (See Figure 2.) Only storage traffic flows over the SAN.
Ethernet is not typically used as the transport for these networks for several reasons. Typically Fibre Channel technology links SAN equipment to servers. So even though SAN cables and Ethernet cables may look the same in some installations, the signals and voltages used on the networks are different.
Modern SAN solutions can be expensive. This is in part because they are typically deployed on an enterprise basis. SANs may also be more expensive because the components are more robust and the equipment is more redundant.
A typical SAN installation is more complex than a simple DAS installation, because it is used to meeting more demanding storage requirements. SAN performance is good. The combination of high-performance networking technology, fast disk drives and the use of dedicated processors all contribute to a speedy solution.
Whereas DAS shares files across a network, SANs share blocks of data. Performance on block-based systems can be significantly faster than on file-based systems where one user must wait until the other user is finished to work on the same file.
It is important that SANs run on dedicated hardware. In a DAS file sharing scenario, if the workstation is busy completing a complicated spreadsheet update, any requests for file sharing are delayed. SANs do not have this problem. They are more flexible. While small SAN installations may not be practical, SANs scale well from midsize to large installations.
SANs are reliable because they almost always incorporate RAID. The block-based nature of SANs makes multiuser access quick and efficient. The systems are specifically designed to allow different users to share data. Broadcasters who want to simultaneously share large amounts of video data in a highly reliable way between multiple users should consider SAN.
Network attached storage
NAS uses the same network for storage data access and other network tasks. The NAS is a dedicated server connected to an Ethernet network. (See Figure 3.) It appears as network drives to a local computer.
The cost of NAS solutions can range from extremely low to very high, depending on the capabilities of the system. In some cases, NAS systems are easy to install.
Manufacturers have spent a lot of time and effort making them interface with just about any computer system. Furthermore, most systems behave like chameleons, making themselves look like native storage to whatever OS is interfacing to them. Inside, the system treats the data in a uniform manner, but at the interface points, the NAS presents the storage data and command interface in a way that the accessing OS understands. These capabilities are even available in low-cost units.
NAS is file-based, so it may suffer a performance hit compared with a SAN's block-based storage system. NAS' performance, however, will be better than DAS' because the server is a dedicated device and does not perform workstation tasks.
NAS solutions can be as unreliable as DAS or as reliable as SAN, depending on the design of the storage media and components. If a single disk is used and that disk fails, a user will lose all of the data. On the other hand, if a user invests in a NAS with RAID disks and multiple controllers, redundancy can be much higher.
Brad Gilmer is president of Gilmer & Associates, executive director of the Advanced Media Workflow Association and executive director of the Video Services Forum.
|Complexity||Low||High||Low to moderate|
|Reliability||Low||Highest||Low to high|
Send questions and comments to: firstname.lastname@example.org