At many customer sites, backup and recovery can be a complex part of automated operations. Over the years, our experience has indicated that a little up-front effort can make new Robot Save customers much more successful at deploying the software. This article is designed to help you start in the right direction.
The real challenge is to build a backup strategy that documents the steps required to restore your data following the destruction of a single object, or an entire system. All types of backups must be automated, and you need to be aware of the amount of system downtime required for backups.
Note: This document was written with the assumption that your backup strategy is complex. If you have no issues with downtime or speed of backups, you are not networked, or you do not use multiple tape drives, contact Technical Support. Our staff can walk you through the process of getting started.
Before you install Robot Save, do the following:
Learn which libraries and directories on your systems are growing/changing dynamically. This allows you to identify your most critical libraries and directories. You should have a plan that backs up the changes to these dynamic libraries and directories daily. During this process, you might identify obsolete libraries you can purge from your system. The following commands can help you identify your critical libraries and directories:
If you own Robot Space, the disk storage management software from HelpSystems, you can generate reports that show you which libraries are changing dynamically. Use the SPCGRTHRPT command to identify the libraries and directories that are changing and are candidates for backup.
If you own Robot Schedule, run the Print Disk Space by Library (PRTDSKSPC) command weekly for all libraries on the system. This report shows only the libraries that are growing the most rapidly. It runs a program that executes a DSPOBJ for *ALL objects in the specified library (or *ALL libraries) to an outfile. Then, it calculates the size of each object and library from that file.
Use the operating system Print System Information (PRTSYSINF) command to print a report summarizing library sizes and libraries not being backed up. It is a large report and you must scan it carefully to locate the information you need.
Once you know which libraries are dynamic, you can build a daily backup routine. Create a Daily backup class and associated backup sets. We recommend that you perform a SAVLIB or SAVCHGOBJ on your dynamic libraries daily. You should execute the SAVSECDTA and SAVCFG commands nightly.
Analyze your business requirements for computer availability. Can you afford for your systems to be down during backups? If so, for how long? Some companies can have their system offline each night for two to three hours; others cannot. Some suggestions about downtime:
Save-while-active has been available in the operating system for several years. Although the name implies saving while an application is active, that's not really the case. Many applications cannot use this feature if they manipulate objects (such as members in a file), allocate objects, or modify object attributes. If you intend to use this feature, test whether it will work with your applications.
Using the save-while-active *SYNCLIB option can possibly reduce your downtime. To use *SYNCLIB:
End the application (make sure your end users are out of the application).
Save the library with save-while-active specified and specify a message queue to be notified when the operation reaches a checkpoint.
When the checkpoint message reaches the message queue, you can restart the application and allow the users back on.
The save should complete and place a copy of the library on tape.
Concurrent versus parallel saves.
Do you have more than one tape drive available for the system you are backing up? If you do, you can use both drives at the same time:
Concurrent saves are a simple way of completing backups sooner by using more than one tape drive.
It is easier to restore from a concurrent save than from a parallel save.
With concurrent saves, you create multiple backup classes and sets, each with a different set of libraries, directories, or special saves. These sets are run at the same time. (Many customers run three saves at a time.)Robot
Robot Save can perform concurrent saves for several years. It's reliable, easy to restore from, and easy to set up.
You can run three or more saves at the same time. (Many customers simulate a SAVLIB *NONSYS writing to multiple tape drives simultaneously in non-restricted state.)
Use parallel saves if one library accounts for a high percentage of your backup downtime. Although IBM promotes parallel saves as the answer to backup problems, they do not clarify the complications. For example:
Parallel saves require a device called a Media Definition. To create this device you must call a system API and Robot Save has a command interface into this API.
You have to create a small CL program around saving this library.
Your restores are more complicated because parallel saves essentially scatter your business data across the tape drives defined to the Media Definition device.
When you attempt to restore at a hot site, or back to your system, all the volumes from the backup must be available. If you are missing one volume, you cannot restore any of the objects from the library.
Restricted state saves cause a lot of downtime on weekends.
The first question to ask is, "Do you really need a weekly backup of your entire system if downtime is an issue?" Some thoughts on weekly backups:
Why back up the operating system each week if it only changes when you add PTF tapes and new releases?
Consider performing a restricted state, full-system save once a month. Run the SAVSECDTA and SAVCFG commands nightly in non-restricted state.
Why back up vendor libraries, such as IBM application libraries, weekly when they do not change? Monthly is good enough.
The only items that truly need to be backed up in a restricted state are the operating system and microcode.
Many customers do the SAVSYS in a restricted state and then run concurrent backups to multiple tape drives at one time. (Some do this as a totally lights-out operation.)
Use the Robot Save Restricted State Utility (RSU) to automate your restricted state activity. That way, no one has to be present to run RSU saves.
Tape drive capacity
Have you evaluated your tape drive capacities recently? You can reduce downtime with a faster backup mechanism. Do you have the capacity to run more than one save at a time? Many customers take advantage of Robot SAVE concurrent saves to help reduce downtime.
High Availability (HA)
Have you looked at HA? If your business truly demands no downtime, you need to look at HA solutions. There are several to choose from.
There are several items to consider about tape management:
How long will tapes be off-site and what retention or expiration dates should you use for daily, weekly, and monthly backups?
Automatic volume movement tracking
Do you want to use Robot Save’s automatic movement feature (some applications call this automatic vaulting)? It keeps track of your volume locations, such as off-site versus on-site.
Data center management
Do you have several systems to manage in one data center? Robot Save’s Data Center Management feature lets you use a centralized scratch pool for multiple systems. For data center management, you need to do the following:
Install and configure Robot Network on every system. Robot Network does not have to be active to use data center management, but the hosts and nodes must be configured properly.
Select one of the systems to be the Data Center Management System (DCMS). That system becomes the focal point for all tape activity.
On the Robot Save System Setup menu, select the Data Center Management option to set up the data center. Before you begin including systems in your data center, run the Import Forecast report. This report identifies all media management conflicts for the systems being added. Address any of the conflicts reported before continuing. If there are no conflicts, you can include the system.
Is tape integrity important to you? Robot Save ’s Data Center Management feature ensures tape integrity across your data center. However, you need to know a few things:
Once Robot Save is installed, it automatically checks all tape activity on your system and makes sure that even tape activity outside of the product is monitored. This means that a HelpSystems program is called each time these processes occur.
You need to verify that these outside processes are not impacted when our program is called. It should not affect your system, but we recommend testing.
The Robot Save Data Center Management feature ensures that one system cannot use another system’s unexpired or reserved tapes.
Automated Media Libraries (AMLs)
Do you plan to use an AML (also known as an automatic tape loader [ATL] or robotic device)? Do you really need one?
Reasons you might want an AML:
They allow random access to volumes. Do you plan to use it this way? Do you have an archival application that can take advantage of this technology?
They allow you to share volumes in a data center setting with multiple systems.
They are more secure for your volumes.
They should be more economical.
They allow you to share tape drives.
Reasons you might not want an AML:
They are more complicated than a stacker unit. If you do not like complications, AML technology may not be for you.
There is very little training available about how to operate these devices.
You need a staff member who is capable of learning new technologies.
It usually takes more than one hardware person to install and configure an AML. Ask your representative if they have another customer they support on this device. You might be the only one.
Tips for using an AML with Robot Save.
Configure your AML through the AML Maintenance panels using the technical documentation available on our web site and the Robot Save User Guide.
Make sure the AML works with your system. We have seen many shops where no one has ever tried to do even a simple SAVLIB to the drive. Test the system before you start working with Robot Save.
Set up Data Center Management if you have multiple systems that will be using the AML.
Always load your tapes into the AML using the Robot Save RBSAMLLRN command (or use the Robot Save menu option to learn the tapes in the unit). Robot Save can populate the scratch pool in our database dynamically by reading this inventory.
Familiar with another tape management package? Here's a Robot Save terminology cross-reference:
|Term||Robot SAVE Term|
|Retention date||Expiration date|
|Automatic vaulting||Move set|
|Generations of tapes||Rotations|
|Tape stacking||Alternate backup set|
|ATL (Automated Tape Loader)||AML (Automated Media Library)|
|Tape drive pooling||Group device|
Keep the following items in mind when using Robot Save:
Data sets are not really used in the IBM i world, but Robot Save can track any media from any save on your system with its ad hoc tape tracking feature.
Cataloging in Robot Save takes place automatically when you run a backup. Volumes are assigned or attached to a backup class and backup set until they expire (if you use a scratch pool).
Complete the following checklist to help you get started using Robot Save :
√ Have initialized media available for the scratch pool.
√ Map out your daily, weekly, and monthly backup and recovery strategies.
√ Analyze your current disk space consumption and clean it up, if needed.
√ Understand your business downtime requirements.
√ If you are you running an IBM i network, install and configure Robot Network. Then, configure the Data Center Management feature in Robot Save.
√ If you want to track media automatically, learn about Robot Save’s Move Set feature.
√ Verify that you have adequate capacity on your tape drives.
√ If you have an AML, test it. Information about configuring the AML to Robot Save can be found in the Robot Save User Guide and the appropriate AML topic under the Self-Service Resources.
√ Plan to use Robot Save’s Restricted State Utility to automate restricted state saves.
Our plan is to make it easier to get started with Robot Save. Backup and recovery can be complicated for many data centers and it does not have to be this way. Customers who plan ahead and are educated in this area, are very successful and perform consistent, automated backups each night. Work with our team and we'll show you the path to automated backups.
Still have questions? We can help. Submit a case to Technical Support.