HP Data Protector Software Cell Manager Planning and Sizing Planning and Configuring a HP Data Protector Software Cell Manager and Managing the Internal Database Growth Table of contents Executive summary............................................................................................................................... 3 Solution description..............................................................................................................................
Specifics for small cells ................................................................................................................... 25 Specifics for large cells ................................................................................................................... 26 Maintenance of DCBF directories ..................................................................................................... 26 Variables influencing DCBF behavior are located in the global options file ..........
Executive summary This white paper provides complementary information on how to plan, size, and maintain a HP Data Protector Cell Manager in a Data Protector cell. Solution description HP Data Protector software is a backup and disaster-recovery software that provides reliable data protection and high accessibility for your fast growing business data. Data Protector offers comprehensive backup and restore functionality specifically tailored for enterprise-wide and distributed environments.
Cell Manager software topology The Cell Manager is the key component of a Data Protector cell. It contains the Data Protector internal database (IDB), and is responsible for the start of backup, restore, copy/consolidation, and media management sessions. Primary Data Protector processes The Data Protector services (Windows) and daemons (UNIX) run on the Cell Manager. Table 1: Data Protector Cell Manager services/daemons Service Data Protector 6.0 ≥Data Protector 6.
Session managers The Cell Manager CRS service listens for session requests and starts the appropriate session managers, which in turn starts the required clients. A dedicated session manager controls the clients for each operation. If a new session is started, an additional session manager session is generated.
IDB architecture The IDB is an internal database located on the Data Protector Cell Manager that keeps information regarding what data is backed up; on which media it resides; the result of backup, restore, copy, object consolidation, and media management sessions; and which devices and libraries are configured. What is the IDB used for? The information stored in the IDB enables the following: • Fast and convenient restore: You are able to browse the files and directories to be restored.
The internal database consists of the following parts: • MMDB (Media Management Database) • CDB (Catalog Database) • DCBF (Detail Catalog Binary Files) • SMBF (Session Messages Binary Files) • SIBF (Server-less Integrations Binary Files) Each of the IDB parts stores certain specific Data Protector information (records), influences the IDB size and growth in different ways, and is located in a separate directory on the Cell Manager.
Catalog Database (CDB) CDB records The Catalog Database stores information about the following: • Backup, restore, copy, object consolidation, and media management sessions This is the copy of the information sent to the Data Protector Monitor window. • Backed up objects, their versions, and object copies • Positions of backed up objects on media For each backed up object, Data Protector stores information about the media and data segments used for the backup.
Table 2 lists the maximum configurable sizes and limits for the HP Data Protector catalog datafiles. Table 2: Catalog datafiles and extensions - maximum settings and limits Data Protector 6.0 ≥ Data Protector 6.1 Size of any IDB datafile (base file or extension file) 2 GB 2 GB Max size of fnames: (base file and all extensions) 32 GB 48 GB Max size of fn1...fn4 (base file and all extensions) 8 GB (16 GB with patch) 16 GB Max size of dirs.
CDB location The CDB is located in the following directory: \db40\datafiles\cdb (Windows Server 2008), \db40\datafiles\cdb (other Windows systems), or /var/opt/omni/server/db40/datafiles/cdb (UNIX systems). When adding an extension file, you can specify a different path. Note: You might consider redirecting the CDB portion of the IDB to a dedicated and fast disk subsystem as this helps to maintain IDB performance in large and/or busy backup environments.
Figure 5 depicts the database table extension wizard. Figure 5: Database table space extension wizard Detail Catalog Binary Files (DCBF) DCBF information The Detail Catalog Binary Files part stores file version information. This is information about backed up files, such as file size, modification time, attributes/protection, and so on. One DC (Detail Catalog) binary file is created for each Data Protector medium used for backup.
Example: Creating new DC directories Create new DC directories to provide additional space for file versions and attributes. If possible, locate them on different disks. Steps 1. In the Context List, click Internal Database. 2. In the Scoping Pane, expand the Usage item. 3. Right-click the Detail Catalog Binary Files and click Add Detail Catalog Directory. 4. In the Allocation sequence text box, specify the order in which Data Protector will choose the DC directory to write information to it. 5.
Session Messages Binary Files (SMBF) SMBF records The Session Messages Binary Files part stores session messages generated during backup, restore, copy, and media management sessions. One binary file is created per session. The files are grouped by year and month. SMBF size and growth The SMBF size depends on the following: • Number of performed sessions • Number of messages in a session One session message occupies approximately 200 bytes on Windows and 130 bytes on UNIX systems.
Regular IDB backups An essential part of the Internal Database configuration is configuring the backup of the IDB itself. Once the IDB backup is performed regularly, the most important preparation for recovery in case of a disaster is done. The IDB recovery is essential for restore of other backed up data in case of a Cell Manager crash. Consider the following when configuring the IDB backup: • Create a separate backup specification for the IDB backup.
IDB notifications Data Protector allows you to send notifications from the Cell Manager when specific events occur. For example, when a backup, copy, or consolidation session is completed, you can send an e-mail with the status of the session. You can set up a notification so that it triggers a report. You can configure notifications using the Data Protector GUI or any Web browser with Java support. Input parameters let you customize notifications. Some input parameters allow multiple selections.
Existing IDB related notifications Health Check Failed Event/notification name HealthCheckFailed A non-zero value returned by the omnihealthcheck command. The command returns zero if the following is true: • The Data Protector services (RDS, CRS, MMD, omnitrig, and Inet) are active. What triggers the notifications? • The Data Protector media management database is consistent. • At least one backup of the IDB exists. For more information on this command, refer to the omnihealthcheck man page.
IDB Space Low Event/notification name IDBSpaceLow One of the following events: • The difference between the maximum and current size of all CDB extension files drops below Filename Tablespace Size Limit Threshold [MB] (default maximum for filenames.dat file, 250 MB) What triggers the notifications? • Free space on any of the disks containing the IDB drops below Disk Free Threshold [MB] (default, 50 MB).
Limitations Internal Database Size CDB DCBF 200 GB–1.6TB 1-50 Stores MMDB 20-50MB Data Protector 6.0 ≥ Data Protector 6.1 CDB max. size 32 GB 48 GB Max # File names Unix (est.) 700 Million 1050 Million Max # File names Windows (est.
Number of Backups Scheduled at One Time The maximum total number of backup sessions running in parallel is 100 on UNIX systems and 60 on Windows systems. The default value is set to five. This can be increased using the MaxBSessions global option. When the number of parallel sessions is larger than 50 (recommended maximum) the probability of hitting one of the system limits on the Cell Manager increases significantly (number of file descriptors, TCP/IP limitations, memory limitations).
IDB key growth factors IDB growth depends on your environment and on Data Protector settings that define how much history and detail you want Data Protector to keep to allow for browsing and search of files. Key factors Impact on IDB growth Details about files and size of the environment Data Protector can keep track of each file and each version of the file. This means that during each backup, one file version record (up to 30 bytes) will be stored to the DCBF part for each backed up file.
The simplified graph in Figure 3 presents the difference in IDB growth when catalog protection is set for a relatively short period of time (one month) versus when the catalog protection is the same as data protection (3 years). Also, the difference in usage of the Log all or Log directories options is shown. The major growth of the IDB lasts until the catalog protection has been reached. After that, the growth is low and determined by the growth of the backup environment.
You can browse directories and files before restoring, and Data Protector can fast position on the tape when restoring a specific file or directory. The information does not occupy much space, since not all file details (file attributes) are logged to the database. Option value: Log directories When this logging level is selected, all detailed information about backed up directories (names, versions, and attributes) is logged to the IDB. You can browse only directories before restoring.
Figure 8: Filesystem backup logging level Influence of catalog protection on IDB The largest part of the Internal Database is proportional to the catalog protection period multiplied by the chosen logging level. The more backups are performed within the catalog protection period, the more data accumulates in the IDB. In other words, it multiplies the data needed to store each file version by as many file versions as are backed up during the catalog protection period.
• None: Provides no protection. • Until: Means that the information in the IDB cannot be overwritten until the specified date. Protection for the information stops at noon on the chosen day. • Days: Means that the information in the IDB cannot be overwritten for the specified number of days. • Weeks: Means that the information in the IDB cannot be overwritten for the specified number of weeks.
Recommended usage of logging level and catalog protection Always set a reasonable level of catalog protection. The only exception is if the Log None option is set (in this case, catalog protection does not apply anyway). If you set the catalog protection to Permanent, the information in the IDB is removed only when media are exported or sessions are deleted. In this case, the size of the IDB grows linearly until the data protection period is reached, even if the number of files in the cell does not change.
Specifics for large cells If the number of files grows into the tens of millions, or there are tens of thousands of files generated on a daily basis, and you use the Log All option, then backup speed and IDB growth will become a problem in a relatively short period of time. In this situation, you have the following options: • Reduce the logging level to the smallest acceptable level. Setting the Log Files option can reduce the IDB size to a third, and setting the Log Directories option to almost a tenth.
3. Balance number Data Protector will select the DCBF directory that contains (proportionally to size limit) the least DCBF file number. Minimum for the following value is selected: (CurrentFiles/MaximumFiles) Variables influencing DCBF behavior are located in the global options file DCDirAllocation=0, 1, 2 Default: 0 (1 for Data Protector 6.1) This global option controls which algorithm will be used to select the directory for the creation of the new DCBF file.
Cell Manager hardware aspects to consider MS Windows, Red Hat Linux, Suse ES, and Sun Solaris10 Organizations are under growing pressure to support non-stop business operations without increasing the IT budget. Given their position at the center of the IT infrastructure, servers play a critical role in determining both the overall availability as well as the total cost of ownership of that infrastructure.
Memory ProLiant servers use a variety of techniques to protect against memory errors. It is advisable to equip the Cell Manager server with Fully-Buffered DIMM technology. To improve memory protection even further, HP introduced Advanced ECC technology. Advanced ECC technology is capable of correcting a multi-bit error that occurs within one dynamic random access memory (DRAM) chip.
For further details, see the technology brief entitled “HP Smart Array Controller Technology,” at http://h20000.www2.hp.com/bc/docs/support/SupportManual/c00687518/c00687518.pdf. HP Smart Array controllers are integrated on ProLiant servers, and support a variety of RAID types including RAID 1+0 and RAID 5. Developed and patented by HP, RAID 6 Advanced Data Guarding (ADG) is further supported on Smart Array controllers.
Cluster support for Data Protector software Cell Manager As a part of its high-availability, Data Protector provides integration with a number of cluster technologies. For details on supported operating system versions, level of cluster support and for supported configurations, refer to the HP Data Protector Product Announcements, Software Notes, and References and the HP Data Protector Platform and Integration Support Matrices at: http://www.hp.com/go/dataprotector.
User security HP Data Protector Users is another security-critical layer of Data Protector. The configuration of users must be carefully planned and tested. Some user rights are very powerful and therefore represent a security issue. For example, the User configuration and Clients configuration user rights enable a user to change the security settings. The Restore to other clients user right is also very powerful, especially if combined with either the Back up as root or Restore as root user rights.
IDB space consumption example Due to the complex nature of IT environments in which backup is a requirement it is quite difficult to find a basic formula allowing to precisely estimating the size of the IDB of a Data Protector Cell Manager.
Table 3: Backup datalist details and backup specifics Sum of Files to backup: 5,0 Million Files per directory: 500 Total Backup Capacity: 10 Objects: 20 % of incremental Backup Change 5,00% Device concurrency: 16 Log level: All Number of copies 0 Data protection: 52 Weeks Catalog protection: 4 Weeks Full backups/week: 1 Incr backups/week: 40 GB Figure 10 is a graphical representation of the above-described scenario.
Using the Internal Database Capacity Planning Tool allows to calculate the estimated IDB growth by making changes to the discussed parameters. Figure 11 shows the IDB growth after changing the Logging level from “None” over “Dir” and “File” and at last “All.” Figure 11: IDB Growth It becomes quite obvious that a careful selection of the datalist configuration parameters/details is recommended to manage the database growth effectively.
On systems running Solaris The Cell Manager must meet the following minimum requirements: • 256 MB of RAM (512 MB recommended) For each parallel backup session, 40 MB of RAM are required and 5–8 MB per data segment size. This means that, for example, if you want to run 60 parallel backup sessions 3 GB of RAM plus 512 MB of data segments are needed.
On systems running Linux The Cell Manager must meet the following minimum requirements: • 256 MB of RAM (512 MB recommended) For each parallel backup session, 40 MB of RAM are required and 5–8 MB per data segment size. This means that, for example, if you want to run 60 parallel backup sessions 3 GB of RAM plus 512 MB for data segments are needed. • 300–425 MB of disk space + approximately 2% of planned data to be backed up (for use by the IDB).