Don't Drop the Ball On Storage Management
By Penny Lunt
August 1998
Have you ever looked around your storage room and realized that youıre filling boxes of RAID or optical disks with images you may never need to look at again? Youıre squandering expensive storage on rarely used items, and pretty soon youıll have to buy more storage cabinets to keep up.
Storage management software helps you make better use of your storage. Is a file never accessed? Load it off onto tape. Are certain files really critical? Back them up and then backup the backup files. Are you making poor use of some of your devices and overworking others? Storage management software juggles your files and devices around so that files are stored in the best places for efficient retrieval and cost savings.
Some imaging applications, including those from Eastman Software, FileNET and OTG, have storage management features such as backup, archival and hierarchical storage management built into them. These are fine for storing your images or electronic documents within these systems, but their capabilities wonıt reach all the word files, email, spreadsheets and others documents that may reside outside these systems. Since imaging rarely takes place in a vacuum, you probably have use for general-purpose storage management software.
Storage management software is becoming more centralized because the market is forcing it to be. Although the world continues to move to distributed computing on NT servers, IT managers are wanting more central control over storage. Thatıs kind of like sending your children to ten different day care centers and expecting to watch over them more closely than if they were all at home with you.
HSM On Hold
A popular storage technique among production imaging and workflow users has been Hierarchical Storage Management. This storage method for large data centers monitors stored data. As a file gets old and less popular, the HSM automatically migrates it to cheaper media. The file typically moves from RAID to optical disk to tape. Each file appears to remain where it was on a hard drive or a file server. A ıstub fileı is put in its place that knows the actual location of the real file and tells the software to go get it when a user tries to open the file.
(Archival software can also migrate data from more expensive to less expensive media in a similar way. But itıs not automatic; the system administrator needs to direct the migration. Also, it doesnıt leave a stub file.)
HSM has never taken off in the general storage management market.
One reason is that for HSM to work, a truly solid backup and disaster recovery system must be in place. ıA lot of companies are still struggling with backup implementation,ı says Mark Nicolett, research director at the Gartner Group (Stamford, CT 650-425-2461). ıThe next step for a company thatıs comfortable with its backup is to use storage analysis tools to understand more about the data out there, who owns it and what applications theyıre using to access it. Then perhaps HSM is a reasonable step.ı
There are other obstacles to HSM. ıNetwork administrators are afraid of it,ı says Edward Cooper, director of corporate communications at Legato. ıThey donıt understand it, they donıt know the value it provides and it brings up political issues theyıre reluctant to buy into.ı
For example, in some implementations of HSM, usersı desktop files are migrated to optical, tape and/or offsite. Suddenly, the administratorıs scope of responsibility and risk increases tremendously. If they lose a userıs data, the user has someone to blame. It can also be frustrating to a user to try to open a file that appears to be on their hard drive and find that it takes three minutes to open because itıs actually in a tape library offsite. Another drawback to HSM is that it requires programming. Also it sees databases as one large file and could potentially migrate them along with other large files, not realizing they need to be accessible.
However, Windows NT 5.0 will include an HSM feature originally created by Eastman Software and now owned by Seagate. That will introduce the world of NT users to HSM and may create some devotees.
ıA lot of times people have been intimidated by HSM,ı says Mike Ivanov, product line manager at Seagate. ıThe initial position for HSM was ıdisk is expensive, so move your data off onto a cheaper medium like optical or tape.ı As disk drive prices came down, people started throwing more disk at the problem.ı
Now thereıs more data than ever on hard drives, and IT administrators are finding that their backup applications canıt back up all the data within their backup window.
That brings about a need for new storage solutions with strong and discriminating backup features. These solutions frequently back up the files that are most important to the company and keeps them online. The less-active files are migrated off to a cheaper medium.
Another plus for HSM is that Microsoft is increasingly trying to win over Unix shops. Lots of Unix users have become familiar with HSM products and will expect to continue using them with NT.
Storage management software provides HSM as well as backup, archival and disaster recovery. These features all depend on each other. HSM without good backup is dysfunctional, as is backup without strong disaster recovery.
Full-Featured Storage
IBMıs Storage Systems Division (San Jose, CA 408-256-2217) began offering HSM for mainframes 20 years ago. Storage was very expensive, so instead of keeping everything on disk, IBM provided HSM. The software would look at the size of files, how old they were and how often they were accessed. If they hadnıt been accessed frequently or recently, they would be moved to less-expensive media like optical and tape.
IBM today offers HSM along with other storage functions for distributed environments in its Adstar Distributed Storage Manager (ADSM). Itıs used in more than 450,000 installations around the world. It provides backup, long-term archiving, HSM and disaster recovery for NT, OS/2, Unix, AS/400, SGI, AIX, HP-UX, Sun Solaris, Macintosh and NetWare systems.
The NT product starts at $795 for basic server ADSM, $4,200 for a network enabler (which provides the ability to back up more than one machine), $100 per seat for single users and $8,495 for HSM, which is sold as an optional feature. The Unix version starts at $1,595 for backup and archiving on a server, $100 per seat and $10,000 for HSM.
One bank customer is storing 12 million check images a day using ADSM. You can set a threshold on ADSMıs HSM feature. If the file server is 80% full, instead of adding new storage to your servers, the system will scan the server to look for the biggest files or the files that havenıt been touched lately.
ıIf you look at statistics, only about 20% of whatıs on your server is accessed on a day-to-day basis,ı says Julie Chiu, ADSM product marketing team leader. ıA lot of your files stay dormant.ı
You can exclude certain files from the HSM operation that you want to remain local. The disaster recovery piece of ADSM takes care of an emergency in which your main file server and your ADSM server containing the backed-up copies of your files both go down in a fire or flood. The disaster recovery lets you take a second copy of your files off-site.
In February, IBM announced enterprise administration -- the ability to manage multiple ADSM servers from one central point on a Web browser. This lets customers keep ADSM servers in London and Japan without adding administrators for the new locations.
Legato (Palo Alto, CA 650-812-6000) has the largest installed base of storage management software -- 3.5 million installations worldwide (some through OEM sales). Legatoıs NetWorker provides backup, recovery, archival and HSM for Windows NT and Unix systems.
Edward Cooper points that unlike some other storage management software providers, Legato doesnıt sell databases, operating systems, storage devices or hardware. ıLegato is 100% focused on protecting data,ı he says. ıWe donıt care what youıve purchased in the past or what youıll purchase in the future. We donıt displace any existing technologies.ı Legato claims to support more devices, operating systems and applications than anyone else in the industry.
NetWorker 5.1 ($15,000 per server and around $100 per seat, depending on the number of seats) backs up file, print, database and application servers and their networked clients. It performs incremental backups each night and full backups once a week. It can backup hundreds of servers and clients and automatically backup your backups in a process called cloning. A $750 disaster recovery feature is a version of Stacıs Replica product for bare-metal recovery of wiped-out servers. This feature recovers operating systems, applications and data on computers quickly. NetWorker for Unix offers HSM as another option.
Legatoıs GEMS product addresses the growing desire of companies to recentralize storage management. GEMS is a Java-based central console for managing all the storage devices in an enterprise.
Seagate Software (Scotts Valley, CA 408-438-6550) recently acquired Eastman Softwareıs Storage Management Group. Seagate sees storage management as four components: data protection, data availability, centralized resource management and policy management. Data protection is provided by Backup Exec for Windows NT, Backup Exec for Netware and a new version for Windows 98 desktops (these products are $695 for a single server and $1,395 for the enterprise).
Data availability is the purview of Client Exec, which lets you copy data from client to server. If a user sitting at their workstation accidentally deletes a file, they can look in the Client Exec Catalog in Windows 95 or 98, see all the different versions of their files and drag and drop them back to their desktop without having to involve an IT administrator.
Seagate made the backup applet in Windows NT, Windows 98 and the forthcoming NT 5.0. They also made the disaster recovery feature for NT 5.0. When Microsoft contracted Eastman Software to provide HSM for Windows NT 5.0, Seagate started OEMing the Eastman HSM product. Within two weeks they acquired that technology from Eastman. So Seagate now owns all the storage on Windows NT 5.0.
The backup, disaster recovery and HSM that are in NT 5.0 are for the host machine only. People who want to back up remote machines, remote clients and the enterprise need to buy the full-fledged ıaftermarketı products. For example, the HSM in Windows NT 5.0, called Remote Storage Server, will be single-tier and will only migrate data to a tape device. Seagateıs aftermarket product -- Storage Migrator -- supports data migration to optical disk and tape across many servers and devices.
Storage Migrator ($2,995) is designed to be easy to use. ıThe last thing you want is users being faced with all kinds of dialogs they have to figure out and loading specific applications on their computers,ı says Jeff Drescher, Seagateıs product manager.
The only thing Storage Migrator does thatıs visible to the user is pop up a file recall notifier if they try to open a file thatıs been recalled. That tells the user that the file they requested is in secondary or tertiary storage and it might take a couple of seconds to recall the file. It reports the estimated time of arrival. Default exclusion rules prevent important items, like your root directory, from being shunted off to tape by the HSM.
Seagateıs centralized resource management product is called ManageExec. It gives you an intelligent view of all your system resources and storage management resources through a Web-based console. It lets you make sure all your important data is getting backed up and that the backup protection policy is working.
The policy management product, Nerve Center, lets you set filters on the messages you get from all your storage devices. Normally the IT person sits there and filters through all of the messages to figure out what really is important. Nerve Center has behavior models that automatically make suggestions for certain problems. If those suggestions fail multiple times, then an administrator is notified. If the Nerve Center determines the systemıs going down, it generates an automatic message in BackupExec to tell it to back up the system.
Stock photography seller Corbis Corp. in Bellevue, WA, uses Seagateıs storage products, including HSM, to manage 40 terabytesı worth of images. Older images are sent to StorageTek tape libraries.
Computer Associates (CA) (Islandia, NY 516-342-5224) bought Cheyenne Software in October 1997. CA was already using Cheyenneıs ARCServe storage product in its Unicenter enterprise computing software. ARCServe 6.5 for Windows NT provides compression of files as it saves them on tape. It enables high-speed backup of volumes by streaming data directly from disk to tape, bypassing the file system. It can backup images while applications are running. It can strip data across tape drives in RAID 0, 1 and 5 methods. A data migration option provides HSM for NT workstations and servers. A disaster recovery feature creates a boot disk set. ARCServe offers centralized administration and a centralized database.
Emassı (Englewood CO 303-792-9700) Amass for Windows NT ($4,000 to $20,000 per license depending on the size of your storage) provides access to files in tape libraries and optical jukeboxes. It makes all the storage devices and media look like one giant hard drive. It supports Sun, IBM, SGI and HP operating systems and many storage devices. It can track 25 million files, and will be capable of more when a 64-bit version comes out. About two-thirds of its 3,000 worldwide customers are imaging installations.
This software doesnıt handle RAID archival (although it can use RAID for caching) and it doesnıt provide HSM or backup. It does keep a RAIMA database of the metadata (information about whatıs in a file, where itıs located and what volume group itıs part of) associated with files stored on optical and tape. It backs up that database. The database is online, so if youıre accessing files and you just want to list a particular directory in the Amass file system, you can do that from the database without going to the media sitting in the library.
A highlight of Amass is its caching capability. If you request a file, it automatically caches related files that youıre likely to request next. For example, if a customer service representative requests one customer record, Amass might cache that customerıs entire file, including images of medical forms, receipts and other documents. That greatly speeds up retrieval for those other documents.
Related Articles:
|
