---

SAN Deployment

Xsan is designed to meet the needs of professional, education, and business users.

In this section:

Storage Consolidation and NAS Replacement
Collaborative Workflow Requiring High Bandwidth
Computational Clusters

Storage Consolidation and NAS Replacement

Businesses in every vertical industry and government agencies are facing constant growth in data and more stringent government regulations for maintaining compliant records. This growth is placing more demand than ever on their storage infrastructures. Previously, most businesses and government agencies relied on a decentralized storage model with many disparate network-attached storage devices deployed throughout the organization—often one for every department or for each network application. This method offers little flexibility, however, and can result in inefficiencies as storage needs grow disproportionately. IT departments must purchase more storage to meet the growing needs of one application, even if existing storage devices may be being under utilized by other applications. This method also results in management inefficiencies, as multiple storage resources are maintained and allocated independently.

Figure 1-2 compares traditional decentralized storage with the data consolidation that SAN technology provides. With traditional centralized storage, each server is connected to a separate storage device. Some storage resources may be under utilized while others are at or near capacity. With SAN technology, data is consolidated into a single storage pool that all computers in the SAN can directly access (through a Fibre Channel switch). This saves money by increasing the efficiency, flexibility, and scalability of an organization’s existing storage resources.

Today, IT departments look at data consolidation as a way to simplify administration and maximize utilization of existing storage resources. Using SAN technology, they can create a consolidated storage pool that can be centrally managed and allocated to any application, anywhere in the organization. By consolidating storage, reducing data duplication, and increasing the flexibility of existing storage hardware, investing in a SAN solution can reduce the storage requirements of organizations and lower IT costs.

• Traditional IT services. The ability to connect many servers to many storage devices, as shown in Figure 1-3, enables IT departments to consolidate data for more flexible and efficient utilization of storage resources. Deploying a dedicated storage network can make network services faster and more reliable, because one server isn’t the single point of access for stored data. Xsan also makes it easy to scale out storage—adding more Xserve RAID systems can increase SAN capacity and performance.

Educational institutions are experiencing the same growth in storage needs that are facing businesses today. Affordable storage and tools for efficiently consolidating and managing storage resources are critical in this area. Most educational institutions have moved beyond student information systems and now maintain all student records on disk, including student-generated data, such as multimedia projects. Many schools use server-based home directories to make students’ work available to them from any computer on the network. These advances place enormous demand on storage resources and file sharing services. Xsan will be used in educational institutions primarily for data consolidation and creating scalable, affordable, and easy to maintain alternatives to decentralized NAS storage devices.

Collaborative Workflow Requiring High Bandwidth

Professionals in the video and audio industries rely heavily on SANs for storing and sharing huge media files. The SAN—not the LAN—is the key to collaboration and efficient workflow for organizations that specialize in professional video and film editing, audio editing, and effects and motion graphics creation.

Capturing video or audio data from tape to digital storage system—the “ingest” process—demands consistent high-bandwidth performance. The SAN must be able to capture media from an ingest station (or multiple ingest stations) accurately and without dropping frames.

Currently, in post-production video and audio workflows, source media is rarely altered. Any alterations, such as color corrections, are noted in the project file and changed media is written as a separate render file. Each editing workstation usually stores these smaller-sized render files on the local drive. During the finishing process, these modifications are transferred on the LAN and merged to create a final master file.

Deployment of Xsan in professional audio and video processing environments eliminates data duplication between systems, which reduces storage requirements. Once data is captured to the SAN, it can be safely and securely stored, and made available to any authorized client workstations without ever being transferred between systems. Each client can work directly on the shared storage device, and multiple individuals can even work on the same file at the same time, streamlining collaboration because there’s no waiting for files to traverse the network. Integration with LDAP directory services enables administrators to impose user-based access controls to secure digital assets. Figure 1-4 shows how Xsan would be deployed in a video processing environment.

Computational Clusters

A computational cluster consists of multiple computers running an application against a single large data set. Computational clusters are increasingly being used in scientific computing, render farms, and other processor-intensive applications. The data set is typically hosted on a file server, and individual cluster nodes access the data using a network file system protocol, such as NFS, which limits scalability and performance. Xsan removes these limitations, providing faster, more scalable data sharing across small (64 nodes or less) and large clusters (more than 64 nodes).

In a small cluster, each node is connected directly to the SAN via Fibre Channel and all nodes can mount the same Xsan volume and read directly from the same files, thereby eliminating data replication.

Large clusters use a variation of NAS-SAN integration. Such clusters typically have one or more head nodes that each manage a portion of the cluster. For example, a large cluster may use one head node for every 25 to 50 cluster nodes. In this case, Xsan handles data management at the head node. Each of the head nodes in the cluster is connected directly to the SAN using a Fibre Channel HBA. These nodes access the data sets directly from the SAN and distribute the data to their nodes using network file system protocols, such as NFS. This method of accessing and distributing data is faster and uses less network bandwidth across the cluster than typical NAS solutions. At the end of the processing run, the processed data is returned to the head nodes, which in turn write directly to the shared SAN volume. Figure 1-5 illustrates the layout of a large computational cluster.

---