iSCI defined.

Over the post decade, network-based storage has become important to the enterprise because of the demands mission-critical applications and their data make on the network. Network storage needs to be available at all times, it needs to support enough bandwidth to service all clients, it needs to be able to shrink or grow with a company and it needs to be easily managed. In summary a good network storage solution needs to meet the following criteria:

** High availability 

** Bandwidth

** Scalability

** Ease of Management

Storage area networks (SAN) and network attached storage (NAS) are excellent solutions that meet some or all of the a network's storage needs.

A SAN is a self-contained network consisting entirely of servers and storage devices. SAN devices are connected by a high-speed, high capacity network.

Currently, Fibre channel is the most popular technology for SAN implementations because of its high-speed non-blocking architecture, its ability to scale from 133 Mbps to one Gbps and beyond, and its flexibility, which allows many different storage protocols (IP,SCSI etc.) to operate over its infrastructure. These combined strengths make fibre channel a fast, stable, and flexible technology for creating storage area networks.

NAS provides a scalable, centralized data repository for servers and users to share information across a network and is flexible enough to support interaction with multiple operating systems simultaneously. Simply adding more capacity to an existing appliance or adding more units to the network increases the amount of available storage.

Therefore, NAS is an excellent choice for networks with diverse computing environments (UNIX, Windows, Linux, Macintosh, etc.) that need large amounts of available shared storage.

Each technology has its strengths and weaknesses. For instance, a SAN is very fast and is highly available, but is very expensive and sometimes difficult to manage. While a NAS is relatively cheap and easy to configure to specific needs, it is not very fast compared to a SAN and has no built in availability features unless additional hardware is purchased. A current trend in network storage called iSCSI will combine the best aspects of both concepts to build a new way of providing fast, highly available, easy to configure, manageable data storage.

iSCSI Defined

Internet SCSI or iSCSI is a technology that is being standardized by the Internet Engineering Task Force (IETF). It is a protocol for sending SCSI commands across an IP network for the purpose of data exchange. iSCSI is used by an application or operating system to directly access SCSI storage devices that are available on the network, which means iSCSI transmissions could use GigE and 10 GigE networks without specialized hardware.

iSCSI works by taking a SCSI command and wrapping it in a TCP/IP pocket. The pocket is sent to the ISCSI storage device where the packets ore assembled and organized in the proper way so the remote SCSI bus can process the commands. The SCSI device will respond and the information is again packaged in a TCP/IP pocket through the iSCSI layer and sent back to the computer that requested it.

SCSI devices are one of the oldest and top performing data storage technologies on the market. A server outfitted with redundant components such as power supplies, clustered servers, and RAID configured hard drives creates an extremely fast and reliable data storage device. However, a server configuration such as this would be limited like a network attached storage device because it would have to transfer files by copying them across the network.

The iSCSI hardware and protocol interface allows an SCSI storage device to be accessed as if it were directly attached to the computer. Communication is performed on a block level instead of a file level therefore removing much of the overhead involved in a traditional network file transfer. In essence, iSCSI allows a computer to communicate with a network based SCSI storage device on a hardware level, but through a TCP/IP interface.

Overall, iSCSI eliminates the need for an expensive high speed network like fibre channel by operating over an existing Ethernet network, it can be used by relatively inexpensive high speed equipment to provide high availability data storage and built in reliability features like RAID, backup severs, and redundant power supplies.

iSCSI Security

The IETF recommends that no iSCSI implementation be configured without security. The ability to compromise an iSCSI storage device could mean substantial loss if the information stored on it is sensitive in nature. There are two different security mechanisms used by iSCSI--authentication and packet protection.

Authentication is required whenever a new connection is made. The sender and receiver negotiate a shared secret through a key exchange. Then the username password authentication is performed by incorporating the username and password into an iSCSI transmission. The iSCSI device verifies the validity of the login before starting the session.

The key exchange is optional, so the username and password could be exchanged in a clear text method, but this is not recommended if there is any possibility of an attacker.

To protect the data in an iSCSI transmission, IPSec is used to provide another layer of authentication as well as ensure the iSCSI transmissions cannot be intercepted or tampered with by using strong authentication and encryption of the information exchanged between the iSCSI client and a server.

The Future of iSCSI

There are some hurdles to overcome before iSCSI can truly be implemented, The main problem for iSCSI is the unpredictable nature of an IP network and the fact that the SCSI hardware standard requires commands to be given in a specific sequence and with certain timing. An IP network is unable to guarantee the orderly and timely delivery of the SCSI commands embedded in iSCSI transmissions since it is a "best effort" technology.

Engineers have developed hardware that overcomes the bursty nature of an Ethernet network. As a result, iSCSI transmissions are buffered and correctly ordered and timed. If you couple this buffering with quality of service, iSCSI is quickly becoming a reality. Currently there are several working drafts for the isCS1 standard, each of which continues to develop specific aspects of the technology. Based on those standards there are several working examples of iSCSI available commercially and as the technology stabilizes into a common standard, it will become a mainstay of network storage solutions.

Conclusion

Current network storage technologies are adequate, but are either very expensive or are limited in performance. iSCSI promises to provide the ability to communicate with a fast, highly available, and relatively inexpensive server at near Gigabit speeds from anywhere on a network. It is also extremely scalable from a tiny network to a global enterprise network, because of the nature of a server cluster and SCSI technology. Once the technical hurdles such as reliable, quality of service and further refinements in the iSCSI hardware and protocol interfaces are overcome, it will revolutionize the data storage infrastructure of modern enterprise networks.

Appendix A-. Abbreviations and acronyms

IP     Internet Protocol
IETF   Internet Engineering Task Force
iSCI   Internet Small Computer System Interface
GigE   Gigabit Ethernet
NAS    Network attached storage
RAID   Redundant array of independent disks
SAN    Storage area network
SCSI   Small Computer System Interface
TCP    Transmission Control Protocol

Appendix B: Sources of additional information:

* iSCSI IETF-Draft

(http://www.ietf.org/internet-, drafts/draft-ietf-ips-iscsi-20.txt)

* Storage Networking Industry Associacition

http://www.snia.org/home)

* Intel: iSCSI

http://www.intel.com/labs/storage/iscsi/)

* iSCSI Review

(http://www.digit-life.com/articles2/iscsi/)

* Diskdrive.com: The home of iSCSI knowledge

(http://www.diskdrive.com/)

* Enterprise Storage Forum

(http://www.enterprisestorageforum.com/)

RELATED ARTICLE: Telecoms and Media Lead in Adoption of Mobile Solutions

Western European companies in telecommunications, media, transport and utilities have a growing interest in mobile solutions, according to IDC's European Vertical Markets survey. Mobility-related projects are currently more common in telecoms and media than in transport and utilities, but the latter are expected to catch up with investments in mobile solutions in 2004. The most common mobile solutions are mobile office applications (currently used by 34% of media and 33% of telecoms companies participating in the survey) and mobile salesforce automation (33% of media and 29% of telcos in the sample). In 2004, 17% of transport respondents plan to adopt solutions for mobile customer support, while 13% of utilities intend to deploy hot desking applications.

'The adoption of mobile solutions in transpot telecommunications, media, and utilities is experiencing steady growth, as the benefits derived from anytime, anywhere access to enterprise solutions become more evident to organizations,' said Claudia Lonardi, senior research analyst with IDC's European Vertical Markets research group. 'Another driving factor is the increasing number of personnel working outside company premises.'

IDC's European Vertical Markets survey reveals that:

* Major mobile solution growth areas in 2004 for all four verticals are XML applications and mobile salesform automation. For transport only, adoption is expected to increase in mobile customer support and mobile office products, while utilities are also looking at implementing hot desking initiatives.

* The range of mobile devices available in the market is ever expanding, but survey results suggest that only a few device types are commonly employed by enterprises in transport, telecoms, media, and utilities for mobile access to corporate information and IT systems. The laptop PC and mobile phone are by far the most widely adopted among mobile devices used by the great majority of companies in the survey sample.

* Regarding implementation strategy, the main message for IT vendors is that, although in general companies are favoring the acquisition of mobile solutions from external providers, in-house custom development still plays a part in the four verticals under analysis and especially in utilities 30% of the sample say they develop mobile applications internally.

* The two major obstacles to the implementation of a mobility-related strategy for all four verticals are that it is not part of the company's overall business strategy and that the ROI proposition is not strong enough to justify the investment.

Alcatel

www.alcatel.com/enterprise