Data Storage & Management: Backing Up to the Future

Data Storage & ManagementAs their school populations grow and data storage/management needs explode, farsighted campus technologists look to advances from the DM vendor community.

I saved my presentation in my personal drive on the server last night, but now I can’t find it. It just seems to be gone. Can you get it back?” “It looks like the mail server is corrupted. When was the last backup?” These sorts of questions, whether from faculty, students, or IT staff, can be an IT nightmare, or they can set in motion a relatively simple recovery. It all depends on how organized and up-to-date your institution’s storage and backup systems are. Advances in data management and backup such as iSCSI Storage Area Networks (SANs), and the move from tape to disk (an affordable, reliable storage medium), make this a good time to examine your own backup and storage strategies.

Washington and Lee:Quick, Reliable Recovery

The ability to recover files quickly and dependably when needed is critical for any backup system. That need for reliability pushed Washington and Lee University (VA) to move to an iSCSI SAN and Symantec’s Backup Exec (originally from Veritas Software; Symantec purchased Veritas in mid 2005).

According to Systems and Network Engineer Jim Bollinger, the school backs up about 2.5TB of data daily from a mix of systems, including Microsoft Windows, Novell NetWare, and Red Hat Linux. The school had been using a disk-to-tape backup system, but struggled with reliability problems. Now, the university depends on an iSCSI SAN storage array with Symantec’s Backup Exec 10.0. The system uses one central console to manage and monitor all of the Backup Exec servers in its data center, backing up first to disk and then duplicating to tape. The tape libraries are in the same location as the servers; the disk arrays in another building. That means a second copy of the data is automatically stored across campus for disaster recovery purposes.

Bollinger says a useful feature in Backup Exec 10.0 is that it allows administrators to define how data should be backed up. That makes it relatively easy to categorize data for backup based on how frequently it changes. For example, the school has a huge GIS (geographic information system) data set that seldom changes—and the entire system is available on the original DVDs. So Bollinger backs up those files infrequently—as opposed to, say, e-mail or Word files that can change daily.

Gonzaga Law School: Disk vs. Tape

Washington and Lee backs up to both tape and disk, but backing up to tape alone is a common solution at many schools. As data volumes grow, however, tape may eventually prove either too slow or unreliable. In its place, disks are growing in popularity because of their increased reliability and dropping price per gigabyte. This appealed to technologists at Gonzaga Law School (WA) who were using a network-attached storage solution to back up nightly to tape via Symantec’s Backup Exec 9.1 software. The school continues to use the Symantec software, but 18 months ago switched to a disk-based backup appliance from Idealstor, according to John Weingarten, the school’s Computer Services coordinator.

iSCSI SANs Change the Storage Picture for Schools

ONE IMPORTANT TECHNOLOGY advance in storage over the past several years is the ability to run Storage Area Networks, or SANs, over Ethernet connections using the iSCSI protocol.

SANs themselves have been around for awhile—the basic idea is to consolidate storage from scattered servers to a single, centrally managed resource. Most schools now use a direct-attached storage system, or DAS, in which storage media are scattered across the network at various locations. But SANs can cut costs and make management far easier. Though traditional SANs use Fibre Channel connectivity, which can be complex to set up and manage, the introduction several years ago of Ethernetbased iSCSI SANs changes all that.

For universities, it’s an important innovation, since many schools have a limited amount to spend on storage, don’t want to make massive investments in infrastructure, and have staff familiar with IP technologies. Those factors play to the strength of an iSCSI-based SAN, which allows campus technologists to create a storage network using well-understood standards like Ethernet and TCP/IP that can run on a wide variety of operating systems.

Washington and Lee University (VA) Systems and Network Engineer Jim Bollinger chose an iSCSI SAN for his storage network for just those reasons. “The advantage of any SAN is that it separates your disk storage from the server,” he explains. That means that if the server hardware changes, say in an upgrade, the storage system remains separate and can simply be unplugged and reattached to the new system, so you’re back up and running easily. In addition, iSCSI SAN “deals in technologies that administrators are already familiar with,” Bollinger says. His storage system communicates using TCP/IP over inexpensive gigabit Ethernet switches.

Bottom line: iSCSI SANs are growing in popularity because there’s no need to build a dedicated and expensive infrastructure; instead, you use the knowledgebase and infrastructure you already have.

Due to speed constraints with the tape system, the school had been running a full backup just once a week, then incremental backups each night, backing up only data that had changed that day. “We had only six hours each night to run a backup, and we were forced to get creative to make our LTO [linear tape open] technology work within this backup window,” Weingarten says. Partial backups can be a problem because in the event of a recovery, if the single full backup is bad, there’s no alternative for restoring current data.

Using Backup Exec and the Idealstor 4 Bay disk-to-disk backup appliance, Gonzaga can now back up all 250GB of data nightly in about five hours; backup times vary based on what sort of data is being accessed by the system, and other factors. After seven backups, Weingarten then overwrites the first backup folder—something he couldn’t do with tape, which allows appending only. For disaster recovery purposes, a set of disks is moved off site once a week. The Idealstor appliance connects to the existing Windows 2003 network via an Ethernet cable and came ready to run out of the box. Idealstor sells its own backup software, but also works with common backup software packages, including ones from Symantec, BakBone Software, and the BrightStor ARCServe Backup from Computer Associates.

IDEALSTOR’s 4 Bay disk-to-disk system

IDEALSTOR’s 4 Bay disk-to-disk system: an efficient backup solution for Gonzaga.

UW-Milwaukee: It’s All About Data Management

One challenge in storing and managing data is keeping it secure while offering appropriate access for those who need it. That’s more a data-management function than a storage function, but the two solutions are often intertwined.

After extensive testing, the University of Wisconsin-Milwaukee is launching the Enterprise Document Management Suite, Xythos’ campuswide enterprise file-storage and -sharing system. In moving to Xythos, the school also switched from a directattached storage (DAS) system to a Fibre Channel SAN running on a Sun Microsystems Solaris system.

Driven by a university initiative called Wise Use of Technology, two years ago a team began assessing UW-Milwaukee’s campuswide networkattached file system. The team eventually concluded that storage should be handled as an enterprise solution, in a secure, centralized manner that can make files available to faculty or students anywhere in the world. That sort of access was especially important because, according to interim CIO Bruce Maas, UW-Milwaukee faculty members (faculty of the larger UW research university) wanted secure online access to papers as they traveled.

There were two main reasons for moving to a new file-storage and datamanagement system for unstructured data, according to Maas. The Xythos system addresses the first: improving file access and security. The second reason is the cost savings engendered by moving from a direct-attached storage system to a SAN. That switch allowed the school to move away from a costly and inefficient storage and backup system of 25 to 40 separate local area networks (LANs) spread across campus, each administered by a different person.

With Xythos to manage file storage, each student will be allocated 250MB of space, each faculty or staff member will receive 1GB, and users can ask for additional space as needed. An awareness campaign is underway for students, and Maas hopes to eventually have all 30,000 students, faculty, and staff using the system. The Xythos system currently has a total storage capacity of 6TB, and could eventually expand to 32TB or greater.

One popular informal storage method that he hopes the new system will replace is that of small USB memory devices, popular with students and faculty for file transport and backup. The devices, however, are highly insecure (especially when used on lab computers and other shared systems), and can introduce viruses. Instead, “we’d like to see students use [the Xythos system] for storing all class-related files and personal Web pages,” Maas says.

The new Digital Research and Instructional Services department at the University of Virginia boasts electronic versions of just about everything a reader might find in an average textbook, but all of the pieces have been marked up by scholars to enhance the original content.

NCCU: Centralizing Storage

Rapid growth can push schools to consider a new data-storage and -management system. At North Carolina Central University, a public liberal arts college in Durham with 8,200 students, the student population has nearly doubled over the past several years, and the school has plans for a major new science center with significant data storage needs. These conditions drove the search for a new backup and storage system that would be less costly to maintain than the original system, and could scale to meet upcoming needs. NCCU was using DAS, with storage media spread throughout the network, attached directly to individual servers. But DAS systems, which are quite common in higher education, can become unwieldy and expensive. For one thing, the processing power of servers is largely wasted when they’re used simply for storage. It can also be difficult to manage, back up, and expand such a distributed storage system. “If you looked at the campus as a whole,” explains Cecile White, NCCU’s IT director, “we were swimming in servers. Everything was attached [to the] DAS.”

So in 2005, NCCU CIO Greg Marrow made the business case for a new system to the administration, focusing on the recent and predicted growth in student numbers. He also described “how directattached storage wasn’t taking advantage of computing power... It made sense to build a centralized solution [instead].”

A RANGE OF STORAGE AND DATA-MANAGEMENT SOLUTIONS
School Washington and Lee University Gonzaga Law School University of Wisconsin-Milwaukee North Carolina Central University
Approximate number of students, faculty, and staff 3,200 725 30,000 9,800
Driver for new storage system Reliability Speed, reliability, ease of recovery, add vs. append ability Security, cost of supporting previous system, anywhere/anytime access to files, remote collaboration abilities Cost, ease of management, staff skill set
Data volume backed up 2.5TB 250GB 80GB 300-500GB
Backup/storage/data management software Symantec Backup Exec 10.0 Symantec Backup Exec 10.0 Xythos Enterprise Document Management Suite for data management; Arkeia for backup EqualLogic PS Series and BakBone NetVault
Backup appliance(s) Overland Storage REO 9000 for disk; 15 assorted tape drives Idealstor 4 Bay backup appliance HP Compaq StorageWorks ESL9326SL enterprise tape library Overland Storage REO 9000 for disk-based backup; ADIC Scalar 24 for tape backup
Operating systems Microsoft Windows, Novell NetWare, Red Hat Linux Microsoft Windows Server 2003 Sun Microsystems Solaris Sun Microsystems Solaris, Novell NetWare, Microsoft Windows, SUSE Linux
Network setup iSCSI SAN Dedicated gigabit backup network Fibre Channel SAN iSCSI SAN
Backup media Disks and tape Ejectable disks Disks and tape Disks and tape

According to White, school technologists and administrators considered a Fibre Channel SAN, but felt the learning curve to implement such a complex system was too steep. Instead, they elected to install an iSCSI SAN, since its use of Ethernet connections and TCP/IP made administration much easier for the relatively small staff, who already understood that technology.

The solution NCCU chose is an EqualLogic PS Series iSCSI SAN, with the capacity to eventually handle a whopping 60TB of data. The school now has two Equal- Logic PS200E systems in its data center, and plans the rollout of a third at the disaster recovery site 15 miles away.

A key driver in the move to a new storage solution was the deployment of a new version of SCT Banner, the school’s enterprise resource planning (ERP) system. While the upgrade gave the school the ability to share student and financial data via an integrated Oracle database, it also made clear the need for a reliable backup system.

What’s Right for You?

Storage software and hardware is one of the fastest-growing areas of technology, and for good reason. Space-intensive audio and video files are growing at a rapid pace, nipping at the heels of existing storage backup and management space and technologies. NCCU’s plan for 60TB of storage for a 8,200-student school—albeit one with rapid growth ahead—indicates where storage is heading. For so many other schools, being smarter about storage will pay off as storage needs swell. It’s time to evaluate your own current and expanding needs.

comments powered by Disqus