Putting the Wraps on Backup Software

Making a shopping list of backup features
and IBM's enterprise backup solution

This month, the Windows NT Magazine Lab concludes its in-depth coverage of backup products for Windows NT with a look at IBM's ADSTAR Distributed Storage Manager (ADSM) 3.1 backup software. We began evaluating backup products in September 1997. Since then, we've looked at solutions to help keep your data safe, and we've learned a lot about NT backup solutions. Based on this experience, we identified factors you'll want to consider before you choose a backup solution for your organization.

The most important factor in determining the appropriate backup software for your environment depends on your network's scope and diversity. First, we assume you use NT as the master operating system (OS) for your network. Second, we assume you use magnetic tape to back up data. Tape is not necessarily better than other media, but it is currently the backup software vendors' preferred medium. If you're considering using hard disks as your storage medium (there are many good reasons for using hard disks, including price and performance benefits), find out whether your backup vendor supports this option. Before we discuss network configurations further, we need to quickly look at backup strategies.

Choosing a Backup Strategy
The basic theories on the best way to oversee the integrity of an organization's critical information involve centralized and decentralized backup strategies. These two strategies are analogous to the ongoing battle between federal and state governments over who is responsible for particular operations.

The centralized model relies on massive networks and hulking robotic storage devices. Several vendors, including HP, IBM, Legato Systems, Software Moguls, and VERITAS Software, offer centralized backup solutions for organizations with more than 500 clients and servers and various OSs. All these vendors migrated their products from UNIX to NT and are experienced in the details, features, and safeguards associated with backing up terabytes of mission-critical information. However, because these vendors' products are built on years of UNIX enterprise-computing experience, configuring the products to run on an NT network can be challenging. Many vendors claim their products run seamlessly in the background once the software is running, and don't require further configuring­but don't believe them. Corporate networks are not static entities. Companies are constantly upgrading software, scrapping old machines and installing new ones, switching departments around, and adding new users to the network. In such a dynamic environment, you will need to constantly adjust your backup configuration.

The decentralized model uses the same technology as the centralized model, but on a smaller scale. Vendors such as Seagate Software, Stac, NovaStor, Barratt Edwards International, and Cheyenne offer small to midsize solutions.

NT Lab's Backup Software Guidelines
Choosing the right backup software for your network requires you to understand the features that backup products offer. We have created the following set of guidelines to help you select backup software.

Environment. The backup software you're considering must provide agents or support for all the OSs (e.g., various UNIX versions, Macintosh, OS/2, DOS, and Windows NT, 95, and 3.1) and software (e.g., Oracle, SAP, SQL Server, and Exchange) associated with your network. If the solution doesn't support the machines and software on your network, it's not a solution.

Interleaving or multiplexing. Interleaving is the ability to write multiple modular data streams from multiple client or server systems to one recording device concurrently. With interleaving, systems administrators can eliminate the network bottleneck that occurs when a tape device has to wait for data coming across the network from one machine. Without interleaving, you might need to run numerous backup operations using numerous recording devices simultaneously to meet the time constraints of your backup window in increasingly full-time networks. With a noninterleaved solution, you might have to track and move more media offsite, which can cause more headaches.

Some vendors don't offer interleaving and instead run tape devices at full throttle to achieve high performance. However, given the recent advancements in capacity and recording speed technology and the likelihood that these advancements will continue into the foreseeable future, investing in a solution that doesn't offer interleaving (especially in an enterprise environment) doesn't make sense.

Device support. Extensive device support, especially for robotic autoloading devices, is a must for enterprise-level backup software. Make sure the backup software you're considering supports your existing and future investment in storage technology.

Client-side compression. Client-side data stream compression lets you send more data to the server for backup in less time. This feature improves the overall efficiency of your backup solutions.

Centralized software distribution. With centralized software distribution in a true enterprise environment, you don't have to visit every client computer to install backup software. You can push the software across the network to each client machine from one location.

Disaster recovery. Effective and efficient disaster recovery must rank high on your shopping list. When you evaluate disaster recovery solutions, ask each vendor how its product handles hard disk partitions and recovers a failed machine; in particular, ask how the product handles NTFS and FAT file systems and their partition sizes during restores. The differences among solutions will surprise you.

Notification. A robust notification feature set including email, pager, fax, and Simple Network Management Protocol (SNMP) is highly desirable. If part of your backup routine fails, you'll want to know about the problem immediately.

Open files. The two strategies for dealing with open files during a backup operation are to skip the open file and back it up later or take a snapshot of the file in whatever transitional state it's in during the backup operation. Both methods are generally satisfactory for most networks, but they might not suffice for 24 * 7 transaction servers where lost data or a corrupt file can mean the loss of thousands of dollars. For these scenarios, you can rely on more sophisticated solutions, such as St. Bernard Software's Open File Manager, that integrate with available backup solutions. Other vendors have robust proprietary methods for handling open files. Be sure you know a company's open file management solution capabilities before you experience a problem.

Media tracking. Your data's integrity is only as good as the media you write it to. Therefore, many backup solutions let you track how many times you write and rewrite to a particular tape.

Restores. Restoring individual files and folders is the backup administrator's most common task. You don't want to have to jump through too many hoops to restore a file or folder after a backup operation.

Multiple runs. Depending on your network configuration, including the number of servers and tape devices, you might want a backup solution that lets you run multiple copies of backup software concurrently. This feature will let you collect information on specific media types without jeopardizing the backup window.

Pre- and post-execution. Pre- and post-execution lets you use a short script to run an application before or after you perform a backup operation. With these scripts, you can perform functions (such as file encryption, virus checking, software compression, data verification, and notification) that you want to include in your backup procedure but that might not be built into your backup solution.

Encryption. Real-time data encryption, especially for transaction servers or financial data, is a must-have feature. However, if you're considering a backup solution that doesn't offer this functionality, you can still include it by running a pre- or post-execution script.

Virus protection. Integrated virus protection isn't necessary, but it's nice to have. You can also use a script file to add this functionality before or after you execute a backup operation.

Hierarchical Storage Management (HSM). Consider the long-term scenario for your storage media's future. Having an HSM solution that integrates with your backup software today can help you better manage your data tomorrow.

Pricing. After you compare the various backup solutions, take a close look at the pricing structure for each package. Features that you might expect to be in the base product, such as disaster recovery, might be add-on agents with an additional price tag. You might have to pay for support for SQL Server, Exchange Server, and other popular applications. Check with the vendor to find out what features come with the base product and what features are add-ons.

Trial version. Try to obtain a trial version of the software before purchasing it to help you identify the product's strengths and weaknesses and evaluate how well the product integrates into your network. Regardless of which package you choose, remember the crux of the backup mission is protecting data, and failure to protect data effectively is just that­failure.

ADSM 3.1
Systems administrators looking for an integrated data management and recovery tool might want to consider the recent release of IBM's ADSM 3.1 for NT. Although IBM is targeting this product toward enterprise-ready businesses, smaller companies that want powerful backup and restore functionality today and HSM capability in the future can readily use ADSM. ADSM for NT doesn't offer native HSM functions; instead, IBM uses Eastman Software's OPEN/stor third-party solution. We will focus on ADSM's backup and restore features. (For information about OPEN/stor, see Bob Chronister, "OPEN/stor 2.0," December 1997).

Installation
Before you install ADSM on your system, IBM recommends that you answer certain questions about your organization's backup procedures, utilization, and requirements. This homework can help you determine where on your network to back up and restore specific data. For example, do you want to back up data from disk to tape or from disk to optical storage, and which data do you want to go to which device? ADSM doesn't interleave data to one tape device; rather, it concurrently writes data from multiple users to different locations, which IBM refers to as storage pools. We followed IBM's recommendations and gathered pertinent data about our backup procedures and the types of data being backed up. We kept this information handy during the installation.

ADSM comes on three CD-ROMs: The first CD-ROM contains server files, and the other two contain desktop clients for NT, UNIX, NetWare, and 27 other platforms. Various multimedia tools guide you through the ADSM installation process. For example, you can use a wizard to install the server portion. The wizard asks you several questions about where you want to write data and files. Unfortunately, IBM needs to better organize these tools to make them truly useful.

When we first opened the software, we questioned ADSM's user-friendliness because the Quick Start manual is more than 50 pages! After you absorb this information, you can explore the 10 large online manuals that describe the administrative and backup clients. The extensive documentation that comes with ADSM is a double-edged sword­it covers many platforms, but you might have difficulty finding answers to your questions. The documentation would be more helpful and less confusing if it were platform specific.

Configuration
After you finish installing ADSM, you must configure the software for your needs and network. IBM includes a Quick Start menu, so you can verify the installation and perform an initial backup to ensure the server and devices are communicating correctly. Unfortunately, we weren't able to find the Quick Start menu and had to search the online Help index to find it. (You can get to the Quick Start menu only by clicking Help from the ADSM main menu.) Although this misdirection might seem like a small gripe, we expected to see the Quick Start menu right after the required system reboot.

After we located the Quick Start menu, we moved to the step that prompted us to start the ADSM Device Driver. Regardless of whether we attempted to start the device driver manually or automatically, we received an error message stating that the system couldn't find the specified device. The ADSM Utilities interface offered little help but suggested that we needed to reboot the system after installing ADSM or the system had a SCSI device conflict. We tested both suggestions several times unsuccessfully. IBM support offered the same two suggestions, so we used the native NT drivers for the remaining tests. Although we didn't encounter any problems with the native NT drivers, we would like to be able to choose between the NT drivers and the ADSM SCSI driver during installation.

Testing
After you configure the server and clients, you can run a backup test to verify that everything is working properly. To back up the server, select Backup Client from the Launch menu in the main ADSM window, choose the drives you want to back up, and click Backup. When we ran the backup test, ADSM ran quickly and efficiently while backing up about 2GB of data. We deleted the files from the server that we had just backed up and ran the restore function in a similar manner. ADSM restored the files quickly and without any difficulties. When you select the restore function, you can restore the files to their original location on the network or specify a different location. This feature is useful and would be nice to have on all backup products.

The Archive feature works similarly to the Backup function, partly because of the same user interface. The archive and retrieve functions worked efficiently and without any difficulties.

IBM recognizes that systems administrators who use its software have certain preferences. For example, when working with a program, some administrators want to use a command-line interface and others prefer to use a Web browser. ADSM includes a command-line client, a GUI client, and a Web browser feature to aid in overall management, administration, and backup.

In addition to the various interfaces, ADSM offers administrators the capability to use any computer on the network to access any other managed node (computer). We were able to run all the ADSM functions from clients throughout the network flawlessly without any noticeable degradation in performance. One area that stood out was the capability to log on to the ADSM server from any computer after we enabled the Web administration client, as you see in Screen 1. An administrator can connect to the ADSM server and change or update all the ADSM options from one screen. We found this interface easier to use than the GUI or command-line clients, because all the options were available through hyperlinks and navigating through all the options was easy.

You can use ADSM to back up local drives and remote nodes (computers) immediately or use the Administrative Client GUI to schedule a backup on these systems at a later time. IBM has prepared the procedures in the documentation for Administrative Client Command Line users but doesn't explain how to set up multiple client backups. We eventually succeeded in backing up multiple clients, but the process was more difficult than it should have been. The procedures need to be clearer and more user-friendly.

ADSM comes with an easy-to-use uninstall option that lets you remove various ADSM components. However, the software lacks a Remove all button, so you can't quickly remove all the components at once.

Evaluating ADSM for Your Needs
Is ADSM the right product for your needs? This decision hinges on your company's priorities, whether you value the security of IBM's reputation, whether you need an expandable backup solution, and how much time and money you're willing to commit to learn the program.

Throughout the tests, we wanted more functionality from ADSM out of the box. IBM claims that ADSM includes many capabilities, such as the Disaster Recovery Management module. However, you must purchase this component (and several others) separately. This approach is akin to a car dealership selling a car with rims and charging you extra for the tires. After we configured ADSM, the software performed well. Unfortunately, ADSM's installation and configuration process can be confusing, which might frustrate users and cause them to become discontent regardless of the product's power and features. If you have the patience and perseverance to plow through ADSM's setup and documentation, you might want to consider it for your environment. Otherwise, wait until IBM shapes this powerful solution into a more user-friendly product before you invest in it.