Introduction to IP SAN

There is no question that the Internet is becoming the predominant media for carrying massive amounts of digital information in an expanding digitized world and in a highly redundant, distributed, and resilient way—much more so than its original designers could have imagined.  Increasingly, the Internet functions as the backbone not only for email (email being the single most used and most mission critical application of all) but also for all sorts of raw data, financial information, knowledge content, and business transactions.  The Internet also carries voice (in the form of Internet-based, long distance telephone), video and, if things go the way I think they will, a virtually boundless digital storage medium for the masses.

As we continue to digitize anything and everything that has that capacity, the requirement for storage, specifically network storage, increases in an ever-growing rate.  Our existing digital storage architecture as it stands today is just not going to cut it in the long term.  The issues relate to the fact that the existing architecture is non-scalable and that additional storage growing based on the old architecture is expensive and complicated to deploy and to maintain in order to keep up with the current growth.  For additional information on this issue, please refer to our white paper on Computer Storage Architecture. 

During the last four years or so, a new development has progressively caught on.  Although it is a bit early to predict that it will be the definitive answer for all networked storage needs, it has at least proven during the last few years to be the first step in finding an answer for the coming generation of networked storage needs.  What is it? Internet Protocol Storage Area Network (IP SAN). Not a new concept, IP SAN is based on the traditional Fiber Channel (FC) SAN. (Also see Computer Storage Architecture.) The one new thing about it is that it uses TCP/IP as its media, rather than FC.  As we all know by now, the Transmission Control Protocol (TCP) and the Internet Protocol (IP) is part of the backbone of the Internet’s suite of communication protocols.  The beauty behind IP SAN is that when it is utilized, networked storage can be available any place TCP/IP goes.  And that is everywhere.

IP SAN has the advantage of not requiring special hardware to deploy it because for the most part its implementation is based entirely on existing technology with a layer of software implementation on both the client and server side.  With the appropriate IP SAN software, a regular server can be turned into an IP SAN server without needing to purchase any additional hardware.

There are IP SAN manufacturers making products specifically for serving the network storage market.  Most of today’s IP SAN manufacturers would like you to think that their hardware implementation is the best option, but I think at the end of the day, software management tools are where the main differences will lie.  There may be differences currently in hardware quality, scalability, and performance, but in time all manufacturers will offer more or less the same product.  Since the components these manufacturers use are pretty much coming from the same sources, there is only so much they can do to make their product different.  And since all of them are using the same iSCSI standard implementation, the communication layer in the software is also the same.  As a result, the manufacturers that can provide more advanced software management tools will be the ones providing greater value.

Traditionally, Small Computer System Interface SCSI is one of the standard interfaces and command sets for transferring data between devices in both internal and external computing devices.  The command set is how a computer “speaks” to the disk storage.

Internet SCSI (iSCSI) uses the same SCSI command set to communicate between the computing devices and storage, via a TCP/IP network.  IP SAN uses TCP as a transport mechanism for storage over Ethernet, and iSCSI encapsulates SCSI commands into TCP packets, thus enabling the transport of I/O block data over IP networks.

Below are the iSCSI and IP SAN architecture block diagrams

iSCSI has proven itself able to handle the demands of high-performance computing applications such as simultaneous visualization and high-volume transactional databases, just like most of the FC SAN can do.  In my opinion, the debate will be resolved if iSCSI can perform well for 90% of what FC SAN is doing currently.  In fact, with the introduction of 10 gigabit Ethernet, I can foresee it performing better than today’s FC SAN.

I am sure though that the FC SAN market will still be around and further developed.  And there are efforts being made currently to allow FC SAN using TCP/IP as the transport layer to overcome its distance limitations so that the technology can be used in a geographically distributed environment, i.e. the Internet.

I am also certain that IP SAN using iSCSI technology is just one of the steps in the evolution to the more robust and transparent networked storage technology that our future demands.  I envision a networked storage technology that will work well with all traditional storage needs and a wave of new-generation mobile applications compatible with multiple wireless standards.

By Benson Yeung, Senior Partner

Back to Top