As storage capacity continues to grow at a compounded annual growth rate that exceeds 60%-90% for Windows NT-demands for external storage under centralized management persist. With these growing needs, continuous availability, ease of management, scalability, and resource sharing-ultimately in a heterogeneous environment-also receive increased emphasis. Two specific architectures are prime candidates to fulfill these requirements: Storage Area Networking (SAN) and Network Attached Storage (NAS).
Neither SAN nor NAS concepts are new, but
their evpolution into the open system environments represents
one of the most important changes affecting the storage industry
in more than a decade. SAN and NAS are about to revolutionize
the industry with ramifications extending beyond storage into
network and systems management and architecture. SAN and NAS no
longer couple storage directly to the server; this architectural
change has important repercussions on the management of data in
an enterprise.
This Paper covers NAS and compares NAS to SAN. Look for other
Papers on SAN at http://www.periconcepts.com/
Network-attached storage (NAS)
Network-attached storage (NAS) is a fully
integrated and dedicated storage solution. It can quickly and
easily attach to a network topology, becoming immediately transparently
available as a network resource for all clients. NAS is platform-
and OS-independent, and appears to the application as another
server. You can plug it in without shutting down the network,
and it requires no changes to existing file servers.
NAS consists of a simplified server that performs only one function-and
performs it well-not having to meet the conflicting requirements
of a general-purpose OS. NAS can only run limited applications,
but its ease-of-use and price/performance are unmatched. Implementations
vary in which storage technologies and devices they support: Some
may restrict themselves to CD-ROM, others to optical technologies,
and still others may provide access to any storage technology
including tape and magnetic disk.
NAS devices contain embedded processors running some sort of OS
or microkernel that understands networking protocols and is optimized
for I/O services. In contrast to the dedicated networks utilized
with the storage area networking (SAN) architecture , NAS devices
directly attach to the standard messaging network, and are addressable
via standard file-system protocols-for example, Network File System
(NFS) or Common Internet File System (CIFS). These devices are
typically optimized for particular tasks, such as file serving,
and tend to accomplish these tasks at very high performance levels.
To applications running on the network, NAS looks like an ordinary
server. To any client, it looks like a large storage device. If
the NAS does little more than allow the network connection, it
is called network-ready storage or a thin server. When NAS provides
additional processing power to perform file and storage management
tasks, it is called a network-attached storage server (NASS).
NASS provides autonomy from a central server or CPU and optimizes
the cost/performance ratio for a wide range of products covering
multiple applications. It consolidates the storage of many application
servers into a single storage service offered by one manageable
pool of physical storage (Figure 1).
NASS may address the interoperability issue among different major
platforms. Interoperability is an important feature that helps
facilitate server consolidation in today's diverse operating environment.
NASS usually takes advantage of NFS, CIFS, and NetWare Core Protocol
(NCP) to be operational under different UNIX, Windows, and Novell
NetWare platforms while providing simultaneous access to the same
files from different platforms. The downside is that NASS systems
move a great deal of data across the messaging network, potentially
degrading overall network performance.
Major participants in the NAS market include Creative Design (CDS),
Meridian, and Axis Communications at the low end; Network Appliance
(NetApp), Unisys, LSI, Procom, and Artecon in the mid-range; and
EMC and Auspex at the high end. Hewlett-Packard (HP) offers NAS
jukebox solutions. Some software vendors offer a complete OS for
NAS that can turn any hardware storage system into a multiprotocol
file server for both UNIX and Windows clients. CrosStor is one
of the leading vendors in this market.
A More Elegant Solution
Traditionally, the approach to solving network storage problems
has been to add new drives and more processing power to available
general-purpose/application servers or to add a new server. This
approach does little to address the real problems of degraded
response time, increased complexity, and availability of data.
The advantages of NAS over conventional server-attached (or bus-attached)
storage are performance and connectivity (see Figure 2). In addition,
if you need an additional server to handle the new storage, a
NAS solution is simpler and cheaper. NAS can produce improved
file-access performance at a substantially lower cost than a general-purpose
network server can. When factoring in the additional cost savings
generated with a simple installation process, which literally
takes minutes instead of hours or days, and ongoing reduced management
costs, NAS is a better, more elegant solution.
New applications have created new requirements for high-speed
transfer of very large files. General-purpose servers and most
OSs were developed for fast processing in multiuser, multitasking
environments and are a poor match for handling large files at
high speeds. Performance problems and traffic bottlenecks develop
when one server transmits data to another, or more accurately,
when one storage device attached to a server transmits data to
another storage device connected to another server on the other
side of the network. NAS is optimized to move the data to users
efficiently without the overhead and complexity of general-purpose
servers. The controller's ability to connect anywhere on the network
lets you balance network performance by placing the storage close
to the users who need the data. This architecture is particularly
effective when bridges, routers, or switches segment the network.
Both storage area networking (SAN) and network-attached storage
(NAS) technologies involve externalizing storage from the server
and adding flexibility to network storage. With SAN technology,
storage devices reside on their own networks with all the flexibility
and performance benefits associated with networking. NAS technology
involves employing a networking interface on storage devices,
making them fully active nodes on the existing network. Both technologies
come with comparative benefits and drawbacks.
Pros and Cons
The advantages of SAN reside in its superior performance, reliability,
and connectivity. SANs offer a high-bandwidth link capable of
growing incrementally and are better suited to transferring very
large blocks of data. By contrast, the bandwidth properties of
the data network characterize NAS networks, which are suited to
efficiently move data in moderate-size segments. SAN delivers
data reliability in a predicted time, while LANs (and NAS) retransmit
data when the network is congested or fails for any reason. SAN
offers a very high level of connectivity via cascading hubs and
switches.
NAS devices typically see storage as files; SANs usually see blocks
of data (see Table 1). This difference represents one of the major
advantages of NAS configurations. Its other big advantages are
ease-of-installation and its ability to offer low-cost entry product
configurations. NAS doesn't require any significant initial investment,
and the technology is available and proven.
NAS is not always a good idea for database applications because
it's file-oriented. It works well, however, for document management
and knowledge management applications. Because NAS does nothing
but hold files for the network, it is flexible; however, it can
also be inefficient at peak times due to network slowdowns. NAS
devices work well for workgroups with high storage demands and
in clustered server environments.
Each architecture has advantages depending on the application
and existing infrastructure. The advantages of NAS include leveraging
the current networking infrastructure and the enormous amount
of development invested in this industry. NAS has a lead over
SAN in heterogeneous data-sharing environments. SANs are at least
two years away from offering this feature on a widespread commercial
basis.
SANs and NAS are likely to coexist for some time and are, in many
respects, complementary technologies. It appears likely that NAS
functions will eventually migrate to the SAN. One way to think
of NAS is as an important evolutionary step on the path to SAN.
By: Farid Neema
PERIPHERAL CONCEPTS, INC.
351 Hitchcock Way, Suite #B-200
Santa Barbara, California, 93105
Tel: (805) 563-9491
fneema@silcom.com
This article was published in the May 1999 issue of Windows NT Magazine