Data Storage is normally categorized as Primary Storage or Secondary Storage, with Primary Storage being that which is accessible by the computer or server CPU, and Secondary Storage is not normally accessible by the CPU, and is used for storing everything from the Operating System and Application Programs to personal user data. It is a non-volatile memory and so maintains its state even after power down.
The main storage area in modern computers which is accessible from the CPU is normally referred to as RAM or Random Access Memory which is a volatile memory used for storing instructions from the CPU. This storage type, often called “Memory” is volatile and is cleared every time the computer is powered off. Most computers and Servers now use a form of RAM known as DRAM (Dynamic RAM), where the computer uses input / output channels to access the memory.
Another storage area in a computer is known as ROM (Read Only Memory) which is primarily used to store a bootstrap file which is commonly known as BIOS (Basic Input Output System), used for booting the computer.
The remainder of this article is going to concentrate on storage of Operating Systems and user programs and data under the sub headings of DAS (Direct Attached Storage), NAS (Network Attached Storage) and SAN (Storage Area Network) utilising Fibre Channel.
Direct Attached Storage refers to a storage medium which is accessible from within the computer or server or from a computer peripheral such as a keyboard. The data goes through the CPU, memory then a bus onto the storage medium such as a Hard Disk Drive. A RAID Controller may be used to organise the data on multiple Hard Disk Drives and provide redundancy and resilience. Files sent to secondary storage systems need to be organised for ease of reading and writing, so a formal file system is needed to do this. Data is organised into folders and files to make location of the data easy. Additional storage can be provided in the form of Flash Drives, Magnetic Tape Drives or ZIP Drives.
Network Attached Storage, or NAS as it is known is a method of storing data across a data network, where the Server communicates with the NAS server which is located remotely within a TCP/IP network. The server which could be Windows, Unix or Linux based, is designed to host common file sharing protocols such as NFS or CIFS. A later development is the use of Appliance servers which have a very basic operating system that is purpose built to support file sharing protocols.
SCSI commands that are block level commands have to be converted to the file protocol level to be passed over the TCP/IP networks and reconverted to block level SCSI commands again when received by the NAS server.
The server just needs a NIC (Network Interface Card) to communicate with the NAS server over the TCP/IP network, which also needs a NIC. NAS is a very common storage option used by small businesses and organisations as well as larger enterprise networks. One of the main problems with NAS is the significant overhead in terms of protocol conversions that must take place as data is sent and received. Fast network connections are needed if data is to be transferred within acceptable time limits and Gigabit Ethernet or even 10 Gigabit Ethernet provide the performance needed for reliable, speedy reading and writing of data.
An alternative to standard Network Attached Storage is iSCSI, which is a method of encapsulating standard SCSI commands within TCP/IP protocols, thus doing away with the need for file protocol conversion. iSCSI was the forerunner to true storage area networks by extending the local bus over a network connection. In order for iSCSI to operate efficiently, it is advisable to run it over a dedicated network or subnet.
True modern Storage Area Networks utilise a dedicated network running the Fibre Channel technology. Fibre channel networks are designed specifically for the transfer and recovery of data from storage servers. Resilience can be provided by means of redundant paths to and from the servers, ensuring no “single point of failure” scenario. The main benefits of Fibre Channel are:
??? Speed of data transfer, which is 2 Gigabits with Fibre Channel 2.
??? The number of devices that can be connected within the network is 127, which is significantly more than SCSI.
??? Distance between the servers and storage facilities can be as much as 10Km, which gives a lot of flexibility and options when deciding where to site the servers and storage facility.
??? Fibre Channel switches allow hundreds of disks to be connected in multiple arrays or clusters.
Fibre Channel has many advantages over other systems such as SCSI and iSCSI and is the storage system of choice for modern medium to large Enterprise networks, but other systems such as traditional Network Attached Storage solutions will continue to be the mainstay for smaller companies and networks for some time.
NAS Systems are manufactured by a number of specialised vendors who provide customers with a wide range of innovative ways to store and access their data.