a data storage device
introduction
What is the thing that holds a data store?
what it is
Three common types, in order of wallet-emptying potential, are DAS, NAS and SAN.
DAS (Direct Attached Storage)
DAS is is permanent storage that is attached to a computer. It is local (inside the computer) as opposed to remote (outside the computer, somewhere on the network). It's a yin and yang thing.
The drawback of an application keeping its own little store of information is that this information is not available to any other applications. A popular way of dealing with heavy demand on an application is to run more copies of the application on more computers. Let's say the business object is a company brochure and you have three web servers dishing out copies of the brochure to web clients. Three copies of the brochure must be stored, one on each web server host. This type of duplication leads to bored webmasters occasionally forgetting to update all the data stores.
Data stored by a computer is written to a disk. If that disk breaks down, and they do have moving parts so they do wear out, the computer is stuffed. One disk can be copied to another for safety, giving a mirror image of the data. If one disk fails, the other takes over. Disk mirroring is one of the functions of the RAID (Replaceable Array of Inexpensive Disks) protocols. It's also slightly misnamed: If you are a fan of the film The Shining and write the word "redrum" on one disk, you don't end up with the word "murder" on the mirrored disk. This kind of insurance against disk breakdown is called fault tolerance.
NAS (Network Attached Storage)
A computer that does nothing but store information is called a storage device. The idea is that you take a computer, chuck out any applications it is running, add a load of disk drives to it and stick it on the network. All the other computers on the network can use it for storage. The business logic computers can do business logic work instead of worrying about storage. If those three web servers in the example above copy the brochure from one central storage location then the webmaster only has to update one data store.
The same safety features thought up for DAS, like RAID, work exactly the same for NAS.
The drawback of a NAS is that data transfer speed is limited to the speed of the network.
SAN (Storage Area Network)
A bunch of storage devices that form their own network are called a SAN (Storage Area Network). Whover decided that a good name for the next evolution of storage was NAS spelled backwards must have been having a laugh.
There are plenty of good things about computers attached to a SAN compared to the original computers-with-their-own-disk-drives setup. A SAN can take care of making backup copies of its information. SAN administration doesn't disrupt any other services. A SAN can be expanded by chucking in more computers and disks. The bad things are it is more complex to manage and more expensive to set up. You have to make sure data can be read and written to the SAN at a cracking pace so other computers don't have to wait for it. Computers in a SAN are connected by Fibre Channel rather than twisted pair ethernet, to avoid crap network transfer speeds. Fibre Channel is a set of standards being developed by ANSI (American National Standards Institute) to make networks fly like shit off a shovel. This means that all an organisation's servers using the SAN must have a superfast fiber optic connection to it.
history
In 1801 Joseph-Marie Jacquard stored loom programs on punched cards. The first big computers in the 1940s also used punched cards.
Magnetic memory was invented after the second World War. Magnetic tape appeared in 1951 and magnetic disks in 1956. The floppy disk appeared in 1970. Every few years since a new I/O method for hard disk drives has been flogged to the general public that is less of a snail than the last method, such as SCSI (1982), IDE (1984), SCSI-2 (1994), ATA-2 (1996), ATA-4 (1998) and so on for ever.
The fault tolerant standards called RAID appeared in 1978.
Optical drives kicked off with the CD in 1977. DVD-ROM drives appeared in 1998.
From a network point of view the first few decades of computer history were selfish. There were only mainframe computers that stored everything, with lots of slave consoles connected to them that did not store anything. All mainframe information was stored on the mainframe's disks. For the first few years of personal computer history they too stored their own information on their own disks.
When networks became the must-have fad in the 1980s computers learned to share. Multiple copies of everything were swept away in a distributed service jihad. Applications started to use a centralised information service instead.
