Storage architectures

Storage architectures can be classified into two types according to the required access, either directly to file or hard drive. Have a view at both types.

  • Hard drive access: Known also as block access, this type of access happens when the client requires direct access to the disk. The file system of the client computer manages the access to disks. There are three possible situations:

1.- Internal disk: The client is accessing directly to its internal disk using the internal buses. If it is a personal computer it would likely use ATA or SATA, and if it is a server it would surely use SCSI.

2.- DAS (Direct attached storage): The client accesses a cabinet of hard drive directly connected to the computer.

3.- SAN (Storage Area Network): The client also accesses a cabinet of hard drives and this time it is not directly connected to the computer but is networked.

  • File access: The client works at the file level, which requests a NAS (Network attached storage) server that manages all the accesses.

In the following figure you can see the four storage architectures discussed:

Storage architectures

DAS (Direct attached storage)

In a DAS storage architecture, the server is connected to a cabinet of hard drives directly through a cable.

The most common situation, with this type of connection, is the use of the SAS protocol, so serial SCSI. Being a server, the connection to the internal disk will also be SCSI (this time parallel). This causes internal and external storage to be handled in the same way by the server.

The use of hard drive cabinets improves scalability and fault tolerance. They are easily scalable structures in which we can add and remove hot disks, in addition to having much more capacity than internal storage. On the other hand, they provide high availability and fault tolerance, through redundancy, RAID levels or software to replicate the data to another cabin in real time.

The drawback is that on many occasions, the storage capacity of these cabinets is underused. Although they can connect to several servers, the number of them that we can connect to is very small, limited by the physical capacity of connections.

SAN (Storage Area Network)

The SAN storage architecture solves the problem of underutilization of the storage capacity that DAS architectures have. Connecting the hard drive cabinets directly to the servers, there are a limited number of connections available and, therefore, the cabinet can only be used with a few servers. To make the hard drive cabinet available to more servers, it seems logical to think of a solution that changes the direct connection to the network connection. In this way, any server connected to the same network as the hard drive cabinet could make use of its storage capacity of it.

Well, the difference between the DAS structure and the SAN structure is that in the first we connect the cabinets directly to the servers, while in the second we connect the cabinets and servers to the same network.

This network to which servers and cabinets are connected is a network dedicated to data storage, in which client computers are not. It is not an IP network, but a SAN network, with its protocols and rules. Therefore, the servers will use different interfaces to connect to the IP network where the client computers are and to the SAN network where the hard drive cabinets are. In the following figure you can see these connections:

SAN Network
SAN Network

NAS (Network attached storage)

In this case, the client works directly with files and has no idea or concern about which hard drives store those files. In these environments, there is a NAS server (Network attached storage) that mounts the File System. The client simply connects to the NAS server and delegates file administration to the NAS File System.


This post is part of the collection “Data Access and Storage Systems”. You can see the index of this collection here