Backup and Recovery Just Got Easier with BRMS V5R3

July 28, 2004 Debbie Saugen

Backup and recovery keeps getting easier as Backup, Recovery and Media Services (BRMS) delivers significant new functions in V5R3 to simplify and automate system backups and recoveries. BRMS is IBM‘s strategic backup and recovery product for iSeries and i5 servers. Packaged as licensed program product 5722BR1 (V5R3), BRMS is a comprehensive tool for managing the backup, archive, and recovery environment for one or more systems in a site, or across a network in which data exchange by tape is required.

BRMS lets you simplify and automate backups as well as manage your tape inventory. It keeps track of what you’ve saved, when you saved it, and where it is saved, so that when recovery is necessary, BRMS restores the correct information from the correct tapes, in the correct sequence.

One of the key highlights of BRMS for V5R3 is the capability to submit restricted-state backups to batch in the controlling subsystem. This is a key feature for backups that users have waited a long time to see happen. Before V5R3, you could run unattended restricted-state saves using BRMS, but it required using the BRMS Console Monitor feature. The Console Monitor feature is an interactive function that emulates batch processing. Console Monitor must be started from the system console and be active anytime a backup is scheduled for the Console Monitor. Console Monitor can be problematic, since backups will fail if the Console Monitor is signed off and not active when the backup is scheduled. In V5R3, Console Monitor is no longer required in order to perform restricted-state backups. You can now simply use the Start Backup Using BRMS (STRBKUBRM) command, specifying *CTLSBS on the submit to batch parameter (SBMJOB). BRMS submits the job to the QBRM/Q1ACTLSBS job queue, which has a job queue entry in the subsystem description of the controlling subsystem. Since you will not have the capability of responding to any messages, any backups you perform, submitted to batch via the controlling subsystem, must be automated. You need to ensure available tapes are mounted or can be mounted for you using a tape library. It’s a good idea to test your restricted-state backup, specifying *CONSOLE for the SBMJOB parameter on the STRBKUBRM command before using the *CTLSBS value. To submit a backup job to the controlling subsystem, you also need to specify *YES on the “allow backup in batch” prompt on the BRMS System Policy. If a problem occurs with the backup and BRMS cannot start the subsystems, you have two options. The first option, which needs to be defined before the backup, is to specify a time limit on the “restricted state time limit” prompt on the BRMS System Policy. The time specified should be that of your backup window. If a problem occurs with the backup, and subsystems are not started within the time limit, the backup job will end and the controlling subsystem will be started. The other option is to select option 21 from the operator panel to start DST, and then select option 14 to “end batch restricted state.”

Another enhancement to restricted-state backups is the ability to run TCP/IP networking in a restricted state. If you are running a restricted-state backup using a shared media inventory, you can now configure BRMS to start a TCP/IP interface to communicate with the remote network systems for selection of tape media. This new function makes the use of a shared media inventory between systems much easier in a restricted-state-backup environment. Previously, you needed to make sure enough media was available and owned by the system you were performing the backup on.

Have you ever wanted to make duplicate copies of your media, but wanted to offload the duplication to another system or partition? BRMS now gives you this capability when you have multiple systems or partitions in a BRMS network and are using shared media. A new from system parameter (FROMSYS) has been added to the Duplicate Media Using BRMS (DUPMEDBRM) command, which allows a system to duplicate media that is owned by a remote system that is specified on the FROMSYS parameter. You still must have two tape devices available on the system to perform the duplication. In addition, when duplicating media in V5R3 using BRMS, any object detail for the duplicated media volumes is maintained.

For those of you Lotus Notes Server users who are already taking advantage of using BRMS to backup your servers while the users are online and active, there are new special values of *LTSOMTONL, *LNKOMTONL, and *LNKOMTLTS. These special values replace the current backup link lists of QIFSXCLLTS, QLTSEXCL, and QLTSXCLONL. By using the new values, you will not have any limitations on the number of omitted links that can be specified, which can be the situation using the current backup link lists.

Since the cost of tape media can be very high, making efficient use of your tape inventory is critical. BRMS has added a new expire media set volumes parameter (EXPSETMED) to the Start Expiration for BRMS (STREXPBRM) command, which will expire media in a set if the volume does not contain active data. You can also use the expire media set volumes parameter (EXPSETMED) when running maintenance using the Start Maintenance Using BRMS (STRMNTBRM) command. The existing normal expiration procedures require that all data on all volumes in the set be expired before any media in the set can expire. This new function can help free up inactive media much sooner.

Another enhancement to the STRMNTBRM command that should allow your daily BRMS maintenance to run faster is a new refree auto-retrieved objects parameter (REFREE). If you do not use the archive function of BRMS or do not retrieve freed objects very often, specifying *NO on the REFREE parameter should improve performance when running your daily maintenance if you are retaining object detail on your saves.

Many of you may have tried using the save-while-active function and were unsuccessful because your applications use commitment control and you could not reach a commit boundary in the allotted time. BRMS in V5R3 provides the user interface to save objects without requiring transactions with pending record changes to reach a commit boundary. There is a new value of *NOCMTBDY that can be specified for the “pending record changes option” under the “save active wait time” attribute of the backup control group. You can edit the attributes of a control group by selecting option 8 to change attributes next to the backup control group you wish to update, on the “work with backup control groups” (WRKCTLGBRM) display. Since objects may be saved with pending transactions, you will need to apply journal changes after restoring the object to reach commit boundaries. This requires using the journal receivers containing information about the pending transactions to apply. With V5R3, BRMS will also track the files and members in user libraries, which are journaled when you save, and assist you when recovering these objects. It is important when specifying *NOCMTBDY for “pending record changes” during the save, that you also specify *YES or *MBR for “retain object detail” on the backup control group attributes. This is required so that BRMS can assist you with the application of journaled changes. When you recover the entire system, the operating system will establish the receiver chains when the journals and journal receivers are restored. You can now use either one of the following commands to assist you with the application of journaled changes:

STRRCYBRM OPTION(*APYJRNCHG) JRN(*JRNLST) 

Or:

STRRCYBRM OPTION(*APYJRNCHG) JRN((library-name/journal-name)) 

Specifying the value *JRNLIST will give you a list of journals in user libraries that BRMS has saved. Specifying library-name/journal-name allows you to specify up to 25 journal names to apply changes to.

There are two more enhancements on the STRRCYBRM command that will help you when recovering independent ASPs or performing a full system recovery. A new option of *ASPDEV and the ASP devices parameter lets you specify which independent auxiliary storage pool device you want to recover. This is useful because the save of the independent ASP could have been saved on the local system, or saved on other systems with the BRMS history information being sent to the local system over the BRMS network. The *RESUME option of the STRRCYBRM command has also been enhanced, allowing system recoveries to resume across jobs. This means that once you start a system recovery, you can signoff, signon, or IPL and still be able to pickup the recovery where you left off.

System consolidation is now much easier for BRMS users in V5R3 when consolidating multiple systems onto a single system. A new value of *MERGE can now be specified on the option parameter of the Initialize BRMS (INZBRM) command, which will merge the BRMS database files to help users with consolidating multiple systems onto a single system.

If you have experienced long install times for the BRMS product because of file conversions in the past, you will see that this is not the case in V5R3. The BRMS install process has been enhanced to perform the file conversions after BRMS has been installed. Also new on the option parameter of the INZBRM command is a new value of *ESTPRDINZ, which allows you to plan for the initialization of the BRMS product. By specifying this option on the INZBRM command after installing the BRMS product, you will receive an estimate of how long it will take to covert the old BRMS database records to their new file formats. To perform the conversions, simply specify *RUNPRDINZ on the option parameter of the INZBRM command.

BRMS is also improving performance by changing how the saved object level detail for backups is structured. If you have lots of files in directories and save the object level detail you should see the amount of storage used be reduced by one-third for the saved detail, the amount of I/O required to save the detail should be reduced by one-half, resulting in faster processing of the saved detail and faster saves of the QUSRBRM library because of the reduced storage.

The BRMS iSeries Navigator client has also added new functions for V5R3 that help to automate backups, to make more efficient use of tape inventory, and to make backup and recovery easier to use. Here are some of the highlights:

• Capability to submit restricted state saves to batch without Console Monitor.

• Reclaim wizard, which allows you to free up tape media that contains expired data or free space and make the media eligible for use again.

• Granular view of backup history for libraries, objects, members, or directories and files when object-level detail is saved.

• Granular search/sort of the backup and recovery log, with capability to search on a specific message.

• Capability to omit libraries, objects, and folders at individual backup policy (control group) level, rather than at the global policy level.

• Controlled shutdown and ending of your TCP/IP or Lotus servers before your backups and starting of your TCP/IP or Lotus servers after the backup.

• Manage Devices function added to Media folder and BRMS task pad for creating, updating, or removing BRMS devices.

• System alias names in lieu of the system name allowed in a BRMS network which can be defined under the Global Policy properties.

• BRMS reports now available in HTML format.

As you can see, BRMS has added some powerful new features in V5R3 for simplifying and managing iSeries and i5 backup and recovery. Keep in mind that BRMS isn’t a replacement for your backup and recovery strategy; rather, it’s a tool that can help you to implement and carry out such a strategy.

For the complete details about BRMS, see Backup, Recovery, and Media Services for iSeries (SC41-5345, in PDF format), as well as the BRMS home page and IBM’s iSeries Information Center.

Debbie Saugen is the technical owner of iSeries and i5 Backup and Recovery in IBM’s Rochester, Minnesota, Development Lab. She is also a senior recovery architect/consultant with IBM Business Continuity and Recovery Services. Debbie enjoys sharing her knowledge by speaking at COMMON, iSeries and i5 Technical Conferences, and Business Continuity and Recovery Services Conferences, and writing for various iSeries and i5 magazines and Web sites. E-mail: dsaugen@us.ibm.com