Getting Started on SAN
As with any new technology, a phased implementation is the best, for several reasons. Primary reasons include risk management, maximized return on investment, improved learning of a new technology, and ease of integration. Allowing the IT staff to incorporate a new technology in a staged and thoughtful way not only enables the SAN architecture, but it also allows its optimization for each unique environment. As the staff gains experience, more components can be added with increased storage management to enjoy more advantages of the architecture. SAN deployment is no riskier than many other IT projects, but like all significant IT projects, it should be carefully managed for best results. Indeed, using SAN for data replication may reduce risk for other critical IT projects such as application development, testing, and migration projects.
Implementation should start with a simple configuration of a single Fibre Channel (FC) hub and a few devices in a local environment. Once this is mastered and extended to greater numbers and types of devices, the next logical progression is to link two FC hubs together with a small switch for additional benefits of distance and logical isolation. Finally, the initial small switch would be tied to a larger switch to gain a more enterprisewide view of storage for improved storage management and cost-effectiveness.
Start with Hubs
The best way to integrate any new technology is to start small, master the technology, and grow. Start the SAN topology in a department with a managed hub connecting the existing server to the new SAN that connects either a RAID or a simple disk array. This will provide a low-cost entry with great flexibility that will be used as a building block for subsequent configurations. A reasonable first step is migrating Small Computer Systems Interface (SCSI) bus devices to the new Fibre Channel way of doing things. Attaching existing SCSI arrays to a protocol converter for Fibre can provide investment protection, improved management, and easy implementation. Even this initial stage of SAN will provide faster access to data, improved scalability, greater distance capability, and improved data availability. Hub implementation will greatly ease the ability to enable a LAN-free/SAN based backup and restore capability with the appropriate application software. This results in a backup procedure that minimizes its impact on the LAN by putting the backup traffic on the SAN.
The hub stage should then be repeated in different departments or facilities as learning improves and the procedures have been tailored to fit the particular installation environment (see Figure 1)5. The end result is a series of SAN hubs installed throughout the enterprise providing improved access, reliability, and performance.
Figure 1 The hub stage.
The hub phase is also known as the SCSI bus migration phase, and it will provide immediate benefits while positioning for subsequent implementations. The money and time spent implementing this phase will be leveraged because the next phase will incorporate the investment made.
Link Hubs to a Fibre Channel Switch
After the initial hubs and the storage system management software have been installed, departments or other logical groups should be linked together. This phase is called the storage infrastructure phase. This phase provides the first glimpse into the broader benefits of the SAN, while still protecting the investment in the hub phase and providing a smooth learning curve to ensure that the enterprise gets the highest return on investment from each phase of the SAN implementation. Linking the hubs with a small switch allows the departmental hubs to be connected with a backup link for higher availability. If one of the primary links between departments has problems, the alternate link can be utilized to provide improved availability. Additionally, it will enable the long-distance connection, up to 10km between two nodes, which provides greater disaster tolerance, as shown in Figure 25.
Figure 2 Long-distance connection between two nodes.
The distance and department segmentation that are provided with a switch will allow the departments to share valuable storage resources and start to develop common management of the storage resource. One of the largest benefits of a SAN is the simple sharing of storage assets, disk arrays, and tape libraries. One significant problem with traditional captive storage architecture is ensuring consistent management of the data in distributed departments. The switch allows hubs to link together while maintaining logical separation, but with fast physical connection. The switch also allows more disaster-tolerant applications because the departmental storage can now be backed up remotely, improving the availability of important data in case of an emergency. The logical separation prevents problems in one department from spilling over to another department.
In the preceding example (see Figure 2), the client now has access to a valuable tape unit for backup via the SAN. It allows a valuable tape library to be used by more clients. This provides better protection for enterprise data with less server overhead, at less cost, and sets the stage for common management of storage assets. This can be done at distances up to 10km to allow groups at a distance to access these resources. It also allows data replication for better disaster tolerance by mirroring data from one site to another via the switch without the distance limitations of SCSI or the overhead of traditional data networks. As the amount of critical data grows and the need for continuous access increases, data replication on the campus with the preceding configuration provides a simple way to protect your operations. This becomes even more feasible as the cost of storage continues to decline.
Data replication, mirroring, and other storage-management tasks require the implementation of the appropriate software to ensure that data integrity is preserved.
Big Switch Phase
Once the hubs and switches are installed together with the storage-management software, it is time for a larger switch. This environment will allow the enterprise to fully recognize the benefits of the SAN, and it will have been implemented in an incremental way with high reliability, legacy equipment utilization, and maximum return on investment of time and capital. It is not necessary to risk the integrity of your data to incorporate a new technology, but it is necessary to manage its introduction. This final stage is referred to as the universal data center because data is now available throughout the enterprise with a variety of applications, locations, and storage devices. During the phased-in introduction, experience has already been gained in the hub-and-switch topologies. Experience gained with storage-management software has allowed consistent storage management even in a diverse environment. The incremental implementation being recommended allows the realization of improved availability/reliability, performance, and configuration flexibility at an attractive cost. The SAN allows the enterprise to maximize the utilization of its LAN, servers, and storage by differentiating the network to provide for its different functions. It allows easier data sharing among different systems, as well as backup that will not impact LAN performance of the clients gaining access to their data. In this final stage, you can find consistent data management, access security, and shared access managed by the SAN in a cost-effective manner that can scale to the largest data centers.
The switch environment, as shown in Figure 3, allows many departments to be linked together with high-capacity dedicated bandwidth for the traffic associated with a large storage network5. For example, large clustered environments benefit from high concurrency found in the large switch.
Figure 3 The switch environment.
Figure 4 shows the three phases of SAN implementation: hubs for the SCSI bus migration phase, small Fibre Channel switches for building a storage infrastructure, and large switches for the universal data center5. By building the SAN in a well-defined and thoughtfully considered plan, each element is more likely to build a low-risk, high-return implementation. SAN provides a superior solution to the concerns of traditional captive storage: reliability/availability, performance, configuration flexibility, and cost. In each area of concern, SAN provides a better way of doing things than traditional captive storage. The SAN architecture is based on Fibre Channel development that started in 1988, and a vast amount of experience has been gained since that time. It is time for your enterprise to enjoy the benefits of SAN technology. This low-risk incremental approach preserves the investment in the equipment from the previous phase and builds on the experience and management techniques used in each successive step. The phased approach outlined in this article will allow a greater opportunity for a successful adoption of the SAN technology, so your enterprise can reap the rewards of better reliability, performance, and configuration flexibility at a more reasonable cost.
Figure 4 The three phases of SAN implementation.