A Storage Area Network can best be thought of as a large pooled storage bucket. As servers or applications require more, the storage is allocated as necessary. A SAN centralizes and shares storage for multiple application servers, and optimizes file serving for end users. Each server participates in sharing the underlying storage resources, but the allocations are dedicated to the individual server or application on a server. This allows the organization to manage storage at a more granular level as well as provide a way to provision storage as needed, rather in large blocks to dedicated server like a traditional direct-attached RAID array.
Additionally, backup resources such as tape drives, tape autoloaders and libraries, or disk-based backup solutions may participate in the SAN. By doing so, these backup resources can offload the transfer of files to the dedicated SAN network and allow the LAN traffic to network clients to be unaffected. In this guide, we'll be addressing basic, essential SAN components and configurations to increase fluency with the technology and facilitate planning.
The following figure illustrates a typical SAN configuration.
As the image above illustrates, there are five basic components in a SAN:
1. Servers
Multiple servers and workstations from different vendors, running different operating systems, can all be connected to a SAN. Systems participating in a SAN usually require a separate high-speed connection to the SAN infrastructure. These connections are made with cards called Host Bus Adapters or "HBAs" for short. There are three basic types of HBAs: Fibre Channel, Serial Attached SCSI, and Ethernet. Within the Fibre Channel technology, there are multiple speeds that can be selected, although most organizations will want to choose the latest technology to get the most benefit from the SAN. Fibre Channel is the traditional implementation for a SAN, as most vendors support this type of connection.
Serial Attached SCSI (SAS) is a newer technology that offers similar benefits to the Fibre Channel implementation, but is typically chosen for smaller implementations. Before choosing a SAS-based SAN implementation, make sure your server vendors and operating systems support a SAS interface.
Ethernet has been a ubiquitous LAN implementation used in organizations of all sizes. With the speed and product availability that gigabit Ethernet offers, iSCSI technology has become a viable alternative to building a Fibre Channel (FC) SAN network infrastructure. Functioning in a similar fashion to Fibre Channel, the iSCSI implementation leverages the expertise and installation of existing gigabit Ethernet infrastructure. The HBAs offered for the iSCSI implementation can offload some of the processing that the operating system would need to do in order to manage the connection to the storage. Future implementations of iSCSI will begin leveraging the 10 gigabit Ethernet infrastructure for even faster SAN connectivity.
2. SAN infrastructure
The SAN infrastructure (also called a 'SAN' fabric) includes the hardware, cabling, and software components that enable data to move into and within the SAN.
For Fibre Channel environments, HBAs and fibre channel switches form the foundation, enabling servers and other storage devices to connect to one another. Switches can detect failed or congested connections and intelligently reroute data to the correct device. When linked together (cascaded), switches increase the number of available SAN connections, which provides greater performance and resilience against individual connection failures. Servers can be connected to a SAN with either a single or dual connection. A dual connection gives you a fallback should one connection fail.
For ISCSI environments, Ethernet adapters (NICs) and Ethernet switches form the foundation, enabling servers and other storage devices to connect to one another. Switches can detect failed or congested connections and intelligently reroute data to the correct device. When linked together (cascaded), switches increase the number of available SAN connections, which provides greater performance and resilience against individual connection failures with either a single or dual connection. A dual connection gives you a fallback should one connection fail. For SAS environments, SAS adapters (HBAs) form the foundation, enabling servers and other storage devices to connect to one another. Servers can be connected to a SAN directly, reducing the cost of having switches. Switches can detect failed or congested connections and intelligently reroute data to the correct device. When linked together (cascaded), switches increase the number of available SAN connections, which provides greater performance and resilience against individual connection failures. Servers can be connected to a SAN with either a single or dual connection. A dual connection gives you a fallback should one connection fail.
3. Disk storage
A disk array can be seen as a centralized storage pool for servers. Data from multiple servers is stored in dedicated areas called logical unit numbers (LUNs), and can be protected against data loss in the event of multiple disk failures using RAID protection. Use of redundant array controllers assures that servers can access their data, even if one controller or network connection fails. The combination of dual ported SAS and SATA drives in the same enclosure reduces cost tremendously while providing the customer the flexibility to use either drive.
In addition to the redundancy that is offered by using redundant disk controllers and RAID, the disk storage units can be added to the SAN infrastructure without taking down the SAN and without disrupting the underlying server operating systems. After new storage has been added, the SAN administrator can allocate the new storage to the serves that require it as necessary, providing better overall utilization of the storage resources.
4. Backup storage
Within a SAN, any disk storage (whether an external disk array or internal to some server) can be backup directly to a tape library or to another backup-only disk array. This provides a fast and dedicated pathway for data backup, and frees the corporate LAN to perform its primary functions with greater efficiency.
Due to their lower network speeds, IP SANs should preferably pass data traffic through a backup server to a direct-attached tape library. Tape-based backup can be combined with disk-based backup to improve recovery processes, optimize slow server backups, and provide affordable tape for long term offsite archival.
5. Management software
Although it is often overlooked, management software is key to SAN considerations. Storage management software should be expected to:
•
Help configure and optimize individual components
•
Monitor the entire network for performance bottlenecks and areas of potential failure
•
Automate time-consuming tasks such as data backup
•
Virtualize storage allowing all available storage to be treated as one virtual pool, regardless of where it's located
•
Replicate data to multiple locations to increase disaster recoverability
There are several categories of storage management software available from HP to help you manage your SAN, including:
•
Storage Resource Management Software (SRM) HP offers array-based replication software for HP StorageWorks XP and EVA disk arrays, and host-based replication software that works with any vendor's storage. This software allows you to take point-in-time copies or snapshots of your data for disaster recovery, testing, application development, reporting, and other uses with HP storage replication solutions.
•
Storage Infrastructure Software HP multipath, failover, volume management, and virtualization software eliminates single points of failure, removes constraints imposed by physical storage boundaries, and simplifies storage management. Find the HP storage infrastructure solution that's right for you.
•
Storage Device Management Software HP storage device management software enables you to maximize your investment in HP StorageWorks disk arrays and fabric switches by providing hardware-specific configuration, performance, and connectivity management capabilities.
Printable version
