Advertisment

IP SAN: The wait and watch game

author-image
DQW Bureau
New Update





Advertisment

The industry consensus is that SMBs will take to it-for cost benefits. But

large enterprises won't rush to replace their fiber channel installations

When the old and complex fiber channel had to die, a new promise-of using

the Internet Prototol in a storage area network over the Gigabit Ethernet-promised

an era of new storage architectures and greater benefits. Not to forget the ooh

aahing over the chea-per new technology!

Even as the storage industry is still polarized into pro and anti-iSCSI

(Internet SCSI) cate-gories, there is one fundamen-tal realization: The

prophecies have largely gone haywire. Yes, IP SAN (Storage Area Network) is

coming, but it is yet to enjoy the market acceptance that many expected.

Enterprises like financial institutions and hospitals, the world over, continue

to bank on fiber chan-nel for high storage per-formance, mission critical

reliability and large complex connectivity.

Advertisment

The hardware components of IP SAN come cheap com-pared to fiber channel. It

opera-tes in a known network, has fewer interoperability issues and is simple.

So why is the adoption rate so slow?

B. Chandrasekhar, country manager with Intransa, feels it is because the

Indian market has always been slow in adopting any new technology. IP SAN is a

relatively new concept. It needs a protocol called iSCSI, which became a global

standard only in 2003. For the last 15 years, people have been using DAS (direct

attached storage) or a fiber channel (FC)-based storage network.

Server vendors were slow to introduce native iSCSI device drivers. Sudhakar

S. Rao, Consulting Director with Hitachi Data Systems India, feels the low rate

of adoption is because the iSCSI is stabilized solely for the Windows and Linux

markets. Besides, there are currently not many IP-based products from the major

storage vendors.

Advertisment

That is a view widely circula-ted. A CIO typically first examines his

differing storage requirements. For storage needs of data-a database, for

example, which is online and needs to be accessed fast-he would, in all

probability, veer towards fiber channel storage. For old data that needs to be

accessed with more time in hand, a more reasonably priced solution like IP NAS

is thought of. George Thomas, country manager with Network Applia-nce, believes

that people will gradually move IP SAN towards more mission-critical

applications.

Country-specific problems could be a hindrance in its adoption too. Says M

Vidyasa-gar, executive VP, Advanced Technology, TCS, "The advan-tage of IP

SAN is that, whatever you would want to retrieve, can be done through the

Internet. This essentially means there is need for very high-speed Inter-net

connectivity at the last mile, and at the enterprise level, which is not

there."

Problems in the way



The implementation of IP SAN needs a significant amo-unt of software

overhead content that has to be executed by the initiating server. "This

software has different eleme-nts, including the requirement for checking lost

packets, interrupt handling on each packet transmitted, integrity check

generation and verifica-tion, and buffer movement of data, among others. In a FC

implementation, this is imple-mented in the hardware," says Sudhakar Rao of

Hitachi Data Systems.

Advertisment

Traditionally, SANs require a separate dedicated infrastru-cture to

interconnect hosts and storage systems. The primary transport protocol for this

interconnection has been. FC networks provide a serial transport for the SCSI

protocol. In addition, IP data transport networks have been built to support the

front end and back end of IP application servers and their associated storage.

"But unlike IP SAN, FC cannot be easily transported over lower bandwidth

long-distance WAN networks in its native form, and therefore, requires special

gateway hardware and protocols. The use of iSCSI over IP networks does not

necessarily replace a FC network but provides a transport for IP-attached hosts

to access fiber channel-based targets," says Sanjay Kharade, principal

consultant, Cisco Systems, India and SAARC.

The major drivers



Economic pressures will force customers to look at alternatives, analysts

say. The price difference between an FC and an IP SAN is fundamentally on three

points: cost of acqui-sition, cost of managing the network and cost of adding

the components. The cost of acqui-sition for FC has been coming down, but is

still double the price of an NIC (Network Interface Card). A gigabit NIC card is

available at $60-70 today. A Fiber HBA (Host BUS Adapter) would not cost less

than $300-400. Intransa expects the IP SAN market to touch about 27 percent of

the global storage market by 2007.

Advertisment

Cisco feels that the industry is positively looking at IP SAN as a solution

and is expecting an increase in adoption in the next one year. What is driving

the IP SAN market is its ability to deliver SAN at a fraction of the cost and

complexity of FC and its superior performance and scalability over NAS.

Initially, very few FC vendors embraced iSCSI, but now they're interested.

"With the global ratification of iSCSI standards, we expect to see a

gradual adoption of IP SANs, particularly with SMB customers who see IP SAN as a

cost-effective alternative," says Kharade.

So vendors are now busy educating prospective buyers, convincing them that IP

network over NAS was a com-promise, and rooting for the power of the IP. But

will it be hot enough? Only 2005 will decide.

Advertisment

Goutam Das

Advertisment