Widely available: shifting storage area networks up a gear

DData, data, everywhere. E-mails, reports, slide presentations, photographs, text files, you name it. The world now produces as much as 2 exabytes (million terabytes) of unique information every year, say researchers at the University of California at Berkeley. That's 250MB for every person on earth.

And now storage — whether temporary or permanent — is critical. Storage requirements are typically governed by company policy, but increasingly they are being driven by industry standards and government legislation. For example, recent EU legislation requires many organisations to store much of their data for seven years.

Since 2000, the trend in the market has been to move storage off the corporate LAN onto a "highly available" SAN. A typical SAN is a pool of storage devices networked via a Fibre Channel switch, which is then attached to the back end of a corporate network through high-speed Host Bus Adaptors that reside on the servers.

A modern SAN comprises three major component areas:

• Fibre Channel Transport
• Physical Storage Systems
• Storage and SAN management software (see Fig.1).

Fibre Channel connectivity allows data to be securely transported at high speed, but is limited in distance in its native format to just 10km. This distance limitation may be overcome through the use of Fibre Channel switches. These can extend the SAN via Protocol Conversion devices which convert the FC to IP, or via intelligent Optical Transport devices that can extend the FC protocol (over WDM or SDH).

Imagine what could happen if a company lost all of its stored data. In severe cases, a business simply would be unable to function. Market analyst IDC estimates that each hour of downtime can cost a substantial company more than EUR50,000. Quite emphatically, time is money and the reputation of a company's brand depends on seamless business continuity.

This is what high-availability revolves around. If a company faces an outage, a highly available storage network prevents network downtime and data loss.

Enterprises requiring high levels of data availability cannot afford to be without a resilient, high-speed inter-networked storage solution. When a significant outages causes a company to lose access to its data, it can be days before normal business operations are restored. And often restoration cannot recover critical data that was in queue.

SANs are often the foundation of a good disaster-recovery plan in that they allow organisations to manage data off the corporate local area network (LAN). SANs enable the best recovery method because they aggregate and store large pools of data that can be synchronously replicated over distance via a number of storage vendor and/or third-party high-availability applications such as: EMC - Symmetrix Remote Data Facility (SRDF); IBM - Peer to Peer Remote Copy; Hitachi Data Systems - TrueCopy; and HP - Data Replication Manager or Veritas NetBackup.

So far, it has not been feasible to transport the Fibre Channel gigabit storage protocols across metro (up to 60km) and regional (200-plus km) networks. Most enterprises have been forced to rely on archaic disaster recovery processes involving manual "tape vaulting". A few large enterprises with substantial IT budgets have implemented remote data replication strategies, but at a high cost.

New optical and storage technologies can transport "storage protocols" across the metro area cost-efficiently. Managing an interconnected storage area network (SAN) is no longer an unfeasible desire. By interconnecting multiple "SAN island" data centres via high-bandwidth and low-latency connections, enterprises can implement a high-availability disaster recovery plan.

The distance limitation of Fibre Channel, the most prevalent network protocol in SANs, has historically been the main obstacle to internetworking SANs over distances above 10km. To alleviate this bottleneck, SAN extension "work-arounds", such as gateways that convert Fibre Channel to ATM or IP, have been implemented. Although these work-arounds have overcome the distance obstacle, it has been at the expense of performance (for example a more than 80% throughput reduction for ATM/STM-1 and more than 40% throughput reduction for ATM/STM-4) and functionality.

New technologies are enabling the extension of Fibre Channel SANs over the wide-area network without functionality and performance degradation.

In the WDM scenario, each wavelength can be added, dropped or passed through the many network nodes. This allows multiple high-speed protocols to be multiplexed and transported over the same fibre, without the need for protocol conversion or compression. Multiple SANs can be connected over a single WDM network. Service changes are made to the WDM nodes without manipulating the fibre, and Fibre Channel signals from the SAN equipment traverse the WDM network at full-rate.

Buffer credit extension technology compensates for Fibre Channel's flow-control methodology that limits Fibre Channel's reach to 10km. This relatively new technology increases the reach to thousands of kilometres, enabling a network of SANs to become a truly remote disaster-recovery solution. An organisation's data can be stored so far away from the organisation's office or offices that any event that affects the offices will not affect the SAN. Added to the fact that SAN disaster recovery can be real-time or nearly immediate, it can be seen why they should be an integral part of high-availability disaster-recovery solutions.

Demand for storage solutions with high-availability disaster-recovery capability will undoubtedly continue to grow. For example, analyst IDC has asserted that 75% of IT spending by 2004 will be on storage.

However, the demand will be coming from a new market sector. SANs used to be the high-end, premium-priced, cutting-edge solutions that only wealthy corporations could afford to build. Now they are rapidly becoming essential IT solutions for any organisation, no matter what the size or location.

Stuart Payne
Director, Channel & Solution Sales
Ciena
[email protected]

Stuart Payne is responsible for developing Ciena's Business Partner Programme, leveraging optical service delivery platforms inherited from ONI's Online range.

Case Study

A reliable data solution was required by an unidentified "billion dollar" international business and IT consulting firm. Its US customers include 43 state and provincial governments, most federal agencies, and hundreds of Fortune 500 companies.

Implementing a local SAN in a data centre could provide OnFiber with shared resources features. However the company wanted to take this concept a step further. Its burning question was: "What would happen if the entire data centre became unavailable?". Realising the potential hazard, the IT firm turned to OnFiber and Ciena.

"Our disaster recovery solution is not complete without a reliable Fibre Channel transport service. OnFiber has the regional presence and, with Ciena, the technology to enable us to implement the solution," said a representative of the IT firm.

The universal interoperability of Fibre Channel technology enabled the client to interconnect multiple vendor storage devices within its data centres, and OnFiber and Ciena between geographically separate data centres.

"Qualified interoperability with major storage vendors have enabled Ciena to provide solutions that extend the SAN across the metro region, driving the implementation of true disaster-recovery applications," said OnFiber.