|
The
Way to SANity
Storage
Area Network (SAN) solutions are becoming a key Enterprise
application and a promising addition to the service portfolio
of many Telecom carriers.
Val
Golan
In
the information age, data has become valuable above all: It
is a resource, but also an output, an intangible asset, and
a wealth to protect. Computer networking revolves around means
to acquire, transmit, exchange and store data. The volume
of data that companies commonly process and store today has
reached numbers that could not be conceptualized a few years
ago. This unprecedented growth has created a huge demand in
data management, and led to the development of Storage Area
Networks, or SANs. Industry analysts consider information
as the lifeblood of organizations and data storage crucial
to an enterprise’s viability. A Gartner Dataquest report published
in Aug. 2002 predicts $21 billion in total worldwide revenue
for external storage, growing to almost $23 billion by 2006.
Of that total, Gartner estimates that worldwide SAN revenue
will comprise $6.5 billion in 2002, growing to $13.5 billion
by 2006, a compound growth rate of 17 percent.
A SAN is dedicated to data storage and
clearly separated from the LAN (Local Area Network) and the
WAN (Wide Area Network). It links the storage-related resources
connected to one or more servers. It is often characterized
by its high interconnection data rates (Gigabits/Second) between
member storage peripherals and by its highly scalable architecture.
Offloading
tasks, such as backup, from LANs and WANs is vital today when
network loads and bandwidth availability are criteria for
measuring performance and even profits. Backup windows have
shrunk dramatically and some environments have entirely eliminated
them, since entire data networks and applications often require
round-the- clock availability.
The
combined factors of great data volume, LAN congestion and
24/7/365 operation have steered the data storage demand toward
the SAN technology.
Evolution
from DAS to SAN
The
first data storage model integrated a number of servers, to
each a dedicated storage system would be attached. Named DAS
for Direct Attached Storage, this technology has physical
limitations that impede system growth and scalability. The
information contained in each server-storage is difficult
to share, and servers that have fully consumed their own storage
cannot use excess capacity available elsewhere. IT administrators
are faced with the time-consuming task of managing these distributed
storage systems individually. Since system growth has to be
carefully planned and over-provisioned, storage costs in the
DAS environment can easily become the costliest item in an
IT budget.
The
emergence of Storage Networking architectures, such as NAS
(Network Attached Storage) and SAN, has brought the promise
of a more dynamic technological concept. Instead of attaching
storage to individual servers, NAS provides file storage directly
to the LAN for use by network applications and users. A SAN
creates a dedicated storage network, separate from the LAN,
using Fibre Channel (FC) or Internet Protocol (IP) switches
that provide block storage for use by servers and workstations.
Although
NAS and SAN are both storage networks, they work with different
protocols and are completely isolated from one another. A
SAN typically interconnects using FC technology, while a NAS
uses IP technology, most often with the Ethernet protocol.
The clear advantage of SAN over NAS is its separation from
the LAN: First, it can transport traffic at speeds much greater
that those supported by the LAN, and more importantly, it
frees the LAN from carrying backup data. This ultimately contributes
to better LAN performance and allows the IT department to
optimize server and network utilization.
SAN
architecture
SANs
can be seen as comprised of two elements: A storage system
and a network. The storage system includes disk storage, tape
libraries, and SAN-management software. The network includes
cabling, switches and different adapters. Host Bus Adapters
are used to connect the hardware (servers and peripherals)
to the network wiring or bus, while switches allow centralized
connection with routing capability, in a similar way to a
LAN switch.
SANs
are comprised of hardware equipment, such as hubs, switches,
bridges, servers and backup devices interconnected through
diverse cabling media. Highly specialized SAN software addresses
the needs to operate the different system components while
enhancing the performance, data availability and manageability
of the SAN.
SANs
can be based on several types of high-speed interfaces. In
fact, many SANs today use a combination of different interfaces.
Currently, Fibre Channel (FC) is the de facto standard in
most SAN environments. FC is an industry-standard interconnection
method and high-performance serial I/O protocol that is media
independent and supports simultaneous transfer of multiple
protocols. Additionally, SCSI is frequently used for sub-interfaces
between internal components of SAN members, such as between
raw storage disks and a RAID (Redundant Array of Inexpensive
Disks) controller.

ă
Dot Hill Systems
Fig. 1 SAN Hardware Architecture
Benefits
of SAN
Transitioning
from DAS to storage networking dramatically enhances the way
an enterprise can manage information. Old limitations are
removed and new techniques enabled. Omar Barraza, Director
of Product Management at Dot Hill Systems, a California-based
Storage Systems vendor, lists the key business benefits for
deploying a SAN as follows:
·
Downtime
Reduction
Using
a network to separate storage and servers allows configuration
changes to occur without affecting other operations. Storage
networking also introduces redundancy that eliminates potential
points of failure between applications and information. Storage
networking reduces both planned and unplanned downtime.
·
Unlimited
Online Scalability
A
storage network eliminates the physical constraints on scalability,
as the storage is now independent of the servers. Information
capacity or processing power can be increased when needed
simply by connecting additional storage systems or application
servers. Adding switches grows the size of the network to
any size needed. Storage networks scale without impacting
operations.
·
Simplified
Management
Consolidating
DAS information onto a storage network greatly minimizes the
number of storage management instances requiring administration.
The resulting reduction in workload on IT staff directly decreases
storage management costs. Industry analysts agree storage
networking and information consolidation lower the total cost
of ownership (TCO) of IT infrastructure.
·
Improved
Capacity Utilization
Moving
storage onto a network enables superior control over capacity
allocation. The exact amount of storage needed by an application
can be assigned, and additional capacity can be added later
if needed. More efficient use of available capacity improves
the return on investment (ROI) of storage assets.
·
Faster
Backup and Restore
Ordinary
backup and restore becomes unpredictable when protecting large
amounts of information due to the volatility of LAN traffic.
Using a storage network eliminates these issues and enables
faster protection and recovery times. It also improves network
application performance by freeing the LAN bandwidth otherwise
used during backup and restore.
·
Disaster
Recovery and Protection
The
switches used within storage networks accommodate very long-distance
connections. The FC switches of a typical SAN can interconnect
across a campus or even across town. IP Switches can be connected
at intercontinental distances. Linking FC and IP switches
into a common storage network is a vital component of disaster
recovery and protection.
·
Broader
Information Sharing
The
“any-to-any” design of a storage network enables an enterprise
to easily and effectively make centrally stored information
available to all users. SAN provides for the sharing of capacity
among servers and workstations, while NAS enables the sharing
of files among applications and users. An enterprise can “mix
and match” SAN and NAS to suit its exact needs.
Additional
techniques, such as “Serverless” backup and server clustering,
further enhance the high availability, reliability and cost-effectiveness
of SANs.
In
“Serverless” backup operations, host computers (servers) do
not handle the backup data, but merely direct and monitor
the backup without actually moving the data. The backup data
is copied directly from disk to tape, or disk to disk, by
“intelligent” storage peripherals. Freed from the routine
data transport burden, server resources can be put to more
productive uses. The backup or tape server becomes a “backup
coordinator” rather than a mere data mover. Serverless backup
takes LAN-free backup a step further since it removes backup
traffic from both the LAN and the backup server.
Server
Clustering refers to grouping together SAN servers for the
purpose of enhancing their performance and/or providing failover
protection in the event that a member server malfunctions.
Uninterrupted and seamless availability of data and applications
during and after a server failure is a primary benefit of
a server cluster architecture within a SAN.
Toward
Convergence: SANs, LANs and WANs
In
today’s global environment, an increasing number of IT operations
have to be performed over geographically dispersed sites.
Data storage frequently involves high-speed data transmission
with FC over existing long-distance fiber infrastructures.
Fiber optics, with its great transport capacity, exceptional
safety, security and immunity properties, can be optimized
into further cost-efficiency with DWDM (Dense Wavelength Division
Multiplexing) systems. Allowing parallel transmission of up
to 32 different channels on a fiber pair, DWDM technology
is emerging in both the core and access components of the
metro network. It provides telecom carriers with greater bandwidth
(multiple Gigabits) with low investment costs (protocol-transparent
DWDM is completely interoperable with legacy equipment). The
benefits of connecting SAN networks over metro DWDM systems
are clear to most players in the IT and Telecom worlds. Benny
Schollin, Managing Director of TSS (Total Storage Solutions)
in Sweden, comments: “In Sweden, where SANs are highly developed
and the fiber optic infrastructure is far-reaching, we have
integrated SAN applications among a variety of companies with
remote sites. We have been using DWDM systems to optimize
on the fiber media and transmit great amounts of data for
storage over distant locations ”.
An
example of such applications is the private telecom network
built by two Northern German publishers to secure data transmission
between their separate headquarters and their common IT site.
The Oldenburg and Bremen publishers need to back-up every
night 500 to 600 gigabits of data, and send to the printers
all the daily newspaper’s articles. The two remotest sites
are located at a distance of 65 km (40 miles), but the PANDATEL
DWDM system would have allowed longer distances. Using a scalable
system, the IT department wanted to keep the option for growth
and extension open.
The
most common way to extend SANs into WANs is to provide Fiber
Channel extension. This allows the interconnection of different
SANs in a point-to-point configuration, using transport mechanisms
such as SONET or ATM. Another commonly used storage
protocol is ESCON (Enterprise System Connection) with data
rates of 200 Mbps. Because of its data rate and protocol-based
technology, ESCON is not easily extended over current available
transport systems such as SONET/SDH rings. DWDM systems provide
channel aggregation for transport regardless of the protocol
or technology, so they will work with any kind of FC extension
as well as point-to-point ESCON.

ă
Dot Hill Systems
Fig.
2 Convergence of SAN, LAN and WAN
With
the growth of packet-based technologies and the cost-effectiveness
of Gigabit Ethernet, an increasing number of optical WANs
are becoming IP-based at the core. Ethernet/IP LANs are easily
extended into these IP WANS, but Fibre Channel, optimal for
storage networks thanks to its high speed and low overhead,
is not IP-based and prevents SANs from reaching into the WAN.
Fibre Channel over IP (FCIP) and Small Computer System Interface
(SCSI) over IP (iSCSI) are two technologies aimed at bridging
the worlds of SANs and WANs. Using these evolving technologies,
SANs can be extended across multiple locations in multi-point
configurations.
When
the SAN meets the WAN, the LAN or the MAN (Metropolitan Area
Network), a whole new paradigm emerges: SAN internetworking.
SAN, LAN, and WAN integration brings data networking and data
storage into a homogeneous point of convergence.
SAN
internetworking allows carriers to expend their portfolio
of Enterprise service offerings with little or no modification
of their WAN infrastructure. Mission-critical customers such
as financial institutions are highly receptive to these new
services in the perspective of secure transmissions and fast
disaster recovery. The current trend toward cheaper and greater
bandwidth, together with the growing demand for readily available,
safe data storage, has made SAN a key application to offer,
at minimal investment costs.
"Carriers
can use DWDM in a shared infrastructure model that provides
unparalleled scalability and redundancy in SAN applications
at a lower total cost of ownership than dark fiber alone,"
said Brian McConnell, director of marketing for WDM products
in Tellabs' optical networking group. "Using this, carriers
can generate new revenue that can quickly add to their bottom
line."
SAN
clearly delivers new opportunities to carriers, and increased
benefits to the Enterprise.
Val Golan is Director of Marketing
and Product Management at PANDATEL, Inc.
|