Storage: it’s one of the essential things in IT, and the demands put on enterprise storage systems are constantly increasing. As these demands increase, it’s important to look at all of the available options and understand what advantages or disadvantages each implementation may hold.
In general there are three types of storage an enterprise may consider: Direct Attached Storage (DAS), Network Attached Storage (NAS), or a Storage Area Network (SAN). Let's start by looking at the simplest example of DAS.
DAS refers to storage that is directly attached to a server—this is usually in the form of SAS, SATA or SCSI disks with an embedded array controller. Redundancy can be obtained via either hardware-level RAID managed by the disk controller or software RAID managed by the operating system. While DAS is often the cheapest option (especially if there is no possibility of increasing storage requirements), there are some disadvantages.
The first problem with DAS is that the maximum disk space is limited; depending on their size, some servers will have as little as two disk slots. With RAID-mirroring enabled this means a maximum disk capacity of around 300 GB, which may seem like a lot but it can all too quickly disappear. If smaller capacity disks were used, then upgrading to increase storage space can be a hassle with the entire system needing to be backed up and redeployed to the new disk set. DAS is also often described as leaving a network with ‘silos of information,’ able to share neither data nor free space with other servers on a network. DAS is not all bad; the greatest advantage of DAS is its high bandwidth and access speed, making it a good candidate for hosting operating systems and swap space.
A SAN describes a data network, which allows clients (usually servers) to connect to centralized storage space at block I/O level via a network. Two main types of SAN infrastructure exist, those being Fibre Channel and iSCSI. Fibre channel networks used for SAN communication will generally be running at 1Gbit, 2Gbit, or 4Gbit with iSCSI running over either 1Gbit or 10Gbit. The use of SANs can bring many advantages to an enterprise. Storage can easily be expanded with all servers sharing the same pool of reserve resources reducing wasted space, while making expansion faster and cheaper. SANs also have many advantages over DAS in the area of disaster recovery. In newer data centres, it’s becoming more usual to have servers booting from a SAN rather than internal DAS; this means that in the event of a hardware failure, a replacement or failover server can boot from the original server's SAN disk, reducing downtime. SANs can enable more effective disaster recovery plans to be formulated; features such as cloning, snapshots, and block-level replication make it possible to have a failover array in a local or remote location. The major disadvantage of a SAN is the initial deployment cost, although many businesses should find a SAN to be more cost effective than DAS over time.
Network Attached Storage can refer to either a server offering file-sharing services to the network or to a dedicated appliance that does nothing but provide data storage, access, and management services to network clients. While SAN communication is at the block level (transmitting raw commands and data), NAS operates at the file level, allowing systems to connect using file-based protocols such as NFS and SMB. While the number of devices connection to a NAS is not limited by physical limitations such as port space, its performance is heavily dependant on the speed of a network and the amount of congestion. Although NAS seems to be less popular than SAN in the enterprise, prices are rapidly falling in the consumer market. Sales are booming, because many new wireless access points and DSL routers include the option to run a NAS server when combined with a USB hard disk.
These are the three main forms of storage to be considered; they are, of course, by no means mutually exclusive, and multiple storage types will likely be used within a corporate network. I would be interested to hear what types of storage readers are using for various applications. Does NFS fulfil your requirements? Has SAN proven to be cost effective and improve manageability? Do you still use DAS as your main storage platform? Leave a comment and let us know.