THE BATTLE OVER HOW TO CONFIGURE
and manage data is heating up. In one corner, there's Network Attached Storage
(NAS) and in the other comer, the Storage Area Network (SAN). Why is this
confrontation happening and what does it mean?Information is the lifeblood of our
economy. The capability of a business to effectively analyze and execute plans
based upon its proprietary information
is one of the defining factors for success in today's marketplace. Now more than
ever, business information needs to be easily accessible, transferable and
scalable. In addition to traditional business needs, other storage-intensive trends
include e -commerce and digital entertainment (movies, music, books, etc.).
With the growing popularity of switched Gigabit Ethernet networks, network
speeds have surpassed disk 1/0. By keeping storage as a component of the
traditional network server, information can only be accessed through the
bottleneck formed by the server's processor and bus, which are busy performing
other application functions in addition to handling requests for data. Another
server could be added with the duplicate information to improve the access, but it
wastes storage space and synchronizing the information is a management
issue. In another scenario, when the server's storage capacity maxes out,
another server has to be added. This results in extra costs associated with
purchasing, configuring, installing and managing an entire server, when all that
was required was some more storage.
NAS and SAN propose different solutions to this problem, each with a distinct
set of advantages and disadvantages. Both offload the storage function from the
traditional server, much in the same way print servers were developed to offload
the print function.
The NAS camp is logically filled with the top players in the server and storage
markets: Compaq, HewlettPackard, IBM, Dell, Seagate, Quantum, Maxtor and
Connex, in addition to a host of network appliance specific vendors. The SAN
camp looks a little stranger, consid
ering its switched configuration. Currently, the Fibre Channel SAN switch
vendors are a cadre of small, new companies: Brocade, Ancor, Vixel, Gadzoox
and McData. Mysteriously, none of the traditional Ethernet switch vendors have
entered this market.
Will one of these two solutions completely dominate the other, or will they stake
out separate market segments and coexist peacefully? The outcome will be
determined by the behavior of the market and the development of faster
connections. NAS has market history on its side. Much like the triumph of
client-server networks over mainframes and Ethernet's victory over anything else,
NAS possesses the deciding factors from those wars--it's distributed, cheap,
easy and compatible.
SAN is more complex. It has the superior configuration and currently the faster
speeds. However, 10Gbit/sec. Ethernet is on the horizon, while Fibre Channel
only has 2Git/sec. to took forward to. The only hope of SAN is the development
of the InfiniBand technology. InfiniBand proposes to completely change the
network, not just the SAN. The mystery of where the traditional switch vendors
went is solved here.
So it all comes down to timing. Cahners In-Stat Group believes the short-term
win is with NAS. It's cheap, easy, compatible, and has a better growth path with
10Gbit/sec. Ethernet than Fibre Channel. In addition, if SAN does end up the
long-term winner, NAS' low cost means the pain of scrapping it will not be as
great. However, depending on the timing of the arrival of 10Gbit/sec. Ethernet and
InfiniBand, NAS could wind up being the long-term winner. After all, the market
rewards the cheap and the fast.
Lauri Vickers is an industry analyst with Cahners In-Stat Group covering the
voice and data communications markets. She can be reached at
/vickers@instat. com
Copyright Electronic News Feb 28, 2000
fred |
