ABSTRACT
SAN (STORAGE AREA NETWORK) TECHNOLOGY AND PROTOCOLS OVERVIEW
DAS vs. NAS vs. SAN
Historically, storage devices, such as disk drives and
backup tapes, were directly attached to a host-hence
the name “direct attached storage” (DAS). This was typi-
cally performed via a SCSI (Small Computer Systems
Interface) parallel bus interface with a speed of up to 320
MBps. This approach of attaching storage devices ema-
nates from the internal computer architecture, which has
obviously reached its limits in several ways. The number of
devices that could be attached to one bus is limited even in
the latest version of the SCSI protocol to only 16 devices,
while the distances are no greater than 15 meters. Sharing
disk or tape drives among multiple hosts was, due to the
architecture of DAS, impossible or required specialized
and typically expensive software or controllers for device
sharing. On the other hand, utilization of the storage spread
across the multiple servers was typically lower than on one
single pool. Necessary expansions of storage volumes and
replacement of the failed hard drives have, in DAS archi-
tecture, frequently generated system downtime. The DAS
architecture is illustrated in Fig. 1. The effort to get better
usage out of storage devices by multiple hosts has gener-
ated specialized devices for shared storage access on the
file level. This architecture is commonly referred as
Network Attached Storage, abbreviated as NAS. NAS
architecture consists of a dedicated device called Filer,
which is actually a stripped-down and optimized host for
very fast network file sharing. Two of the most typically
supported file systems on Filers are NFS (Network File
System) for the UNIX world and CIFS (Common Internet
File System) for the Microsoft world. While the NAS
solution has its simplicity in maintenance and installation
as its main advantage, its main drawback is limited file and
operating system support or support of future new file
systems. The NAS architecture is illustrated in Fig. 2.