With some parameterts a backup is simulated and as a result you get the read performance which would be achieved during the backup. The backup will be done to a null device and will not require a configuration in DP beforehand.
In addition statistical information is collected with -profile. Run Time ……….. Directories …….. Note: The tables are examples and must be viewed individually depending on the unique requirements of the organization. Backups are rarely carried out in the backup environment for 24 hours without interruption, because during the day there is often the need to give priority to production applications.
Backups therefore usually take place outside of business hours, in a so-called backup time window. To determine the required hardware, it is essential to know how much time is available for the backup and the volume of data to be backed up within this time. For example, if the time is not sufficient for backup onto a tape device, the solution must be planned as larger or alternatives such as the snapshot-based backup on memory level Zero Downtime Backup must be considered.
The HP Storage Sizer provides support in determining the required hardware. It plans the right libraries and appropriate StoreOnce system. The Sizer tool is available to partners of HP. Tip: For successful implementation of a backup and recovery environment, an examination of the data in external offices is just as important as information regarding the data in the headquarters.
For example, if the data of the external office are to be replicated as deduplicated in the headquarters, the necessary requirements with regard to required capacities for StoreOnce systems and libraries must be met in the headquarters right at the start of the project. For backup jobs running in parallel, another 40 MB are required assuming 10 backups running simultaneously, this means MB , 64 MB for the installation server, MB for the user interface and 64 MB MB recommended for the media agent.
In total and rounded up, this totals a requirement of 1. Initially, there is no requirement in the Quickspecs regarding the number of processors.
If the StoreOnce software store in HP Data Protector is used, the recommendation is to deploy another 4 to 6 processor cores and 6 GB main memory exact formula can be found in the Quickspecs.
As a general minimum recommendation and as preparation for any future functions in the software, the backup server should have at least 8 processor cores and at least 8 GB main memory. Experience has shown that upgrading an existing server rarely makes good sense; instead, acquisition of the current HP ProLiant server model in each case is recommended.
Tip: During planning of the backup server, attention should be paid to ensuring that the server is procured with more computing performance and main memory than required. This allows, within the framework of the lifecycle, new functions in HP Data Protector that have higher hardware requirements can be integrated for example: integration of the StoreOnce software store.
Which DP components are required and what is their purpose? On installation of the Cell Server, you are asked which components are to be installed on the system. As a rule, the proposed selection is the right one for the Cell Server, supplemented by the Disaster Recovery component. This applies to HP Data Protector versions 7. X and 8.
Note: Before installation of the corresponding components, the support matrices must be checked with regard to compatibility of the versions. X or higher. The language integrations vary in the currently supported HP Data Protector versions. Tip: Only the required components are installed on the Cell Server. Attention should be paid to ensuring that the Cell Server is only used for backup and recovery tasks. The installation of other applications on the backup server endangers the calculated or specified backup and recovery times.
What is a patch bundle and what order is required for updates? No software is free of errors, which is why updates for HP Data Protector are made available at regular intervals to rectify known problems or integrate new functions into the software, for example support for a new operating system.
Note: A valid maintenance contract is required to download the updates from the HP website and their installation. A patch bundle is a collection of different related updates.
The installation of the patch bundle is automated. Under certain circumstances, individual updates that are released later require the installation of a patch bundle. It is therefore recommended that you read the corresponding patch description before installation.
Tip: The updates should be installed directly after installation of the Cell Server and before the clients to be backed up are installed. When downloading and saving the updates, it helps to extend the file name with a description for example: DA or MA for disk agent or media agent.
Tip: A check of the name resolution for the server to be set up should be carried out before installation of the software. This prevents later errors in the backup and recovery environment. HP Data Protector software is installed on the backup clients via its user interface.
This accelerates the installation of the software on the clients. A client receives only the required components. As a rule, the disk agent and the automatic disaster recovery module are sufficient for backup of the file systems.
The integration agents are only installed if backups of applications such as Microsoft Exchange or Oracle are desired.
If an integration agent is installed but not used, Data Protector still requests a license. If no more licenses are available, a missing license is indicated in the list of licenses of Data Protector. Following any changes to the file, the services have to be restarted on the Cell Server. Tip: With the introduction of version 8.
The values for the individual parameters can deviate from the suggestions in each customer situation and must be regarded separately for each environment in all cases. The parameters that have often proved to be effective in practice are listed below in no particular order. They were regarded as expedient as a result of experience gained in different installations. This limitation was already raised to the value with HP Data Protector 8. Adapting the value avoids, among other things, mount requests.
It is useful as information in a disaster recovery scenario. If backups are running at this point in time, the time for maintenance can be shifted accordingly. If backups are running at this point in time, the time for the daily checks can be shifted accordingly. The default value is 0. This option is very helpful for transaction backups.
The default value is 30 minutes. When the time has elapsed, the backup is aborted. The default value is 60 minutes. This option is very helpful for transaction backups onto a hard disk area. The default value is objects. This limitation was raised to the value with HP Data Protector 8. This value was raised to the value 4 with HP Data Protector 8. As default for LTO media, the number of overwrite operations is and for the validity of the media 36 months.
Tip: Changes to the file should be made directly after the installation, before the environment is deployed. When updates are loaded, it should be checked in the corresponding description of the patches whether variables have to be changed to achieve certain characteristics of the cell or whether they are no longer required. In the case of major versions of HP Data Protector, an adjustment and extension of the variables take place during the installation procedure. Depending on the operating system, changes in the file can have a positive effect on the performance of the backups and thus shorten the backup or recovery times.
Changes to the file are used for the next backup or recovery operation; as a rule, it is not required to restart the services. The file should be created not only on the Cell Server, but also on each media agent and disk agent client to be backed up. For Unix, the parameters recommend below can differ.
The values for the individual parameters can deviate from the suggestions in each customer situation and must be regarded separately for each environment but do not have to be complete in all cases. They were regarded as faster, as a result of experience gained in different installations.
As default, the parameter is not available. The default value is 1. The default value is seconds. The default value is 0 and the hard disk runs full.
It was introduced in DP 6. Under certain circumstances, the default value 0 means that the required ports have be added manually on the client. In the case of major versions of HP Data Protector, there is no adjustment or extension of the variables during the installation procedure. Adaptations must be carried out manually after the installation. The segments here are not written onto a device in one operation, rather in smaller units as blocks.
This value applies to DP 7. X; for DP 8. X, the default is kB. When an iSCSI tape device is used in the Windows environment virtual or physical , for example, the registry must be modified first to enable a larger block size. As this is not possible during the disaster recovery operation, because the connected devices have to be addressed with a block size of 64 K during the backup itself.
This applies to the Cell Server and possibly also to the media agent. Therefore, there must be a check beforehand as to what size is supported by the environment, because if the selected block size is too large, HP Data Protector is unable to restore data from a device configured in this way.
Also bear in mind that the block size must be changed before formatting the tapes, as the block size is written in the medium header of the tape so that HP Data Protector recognizes the block size to be used on loading the tape.
If the block size on the tape and the device are different, error messages can occur. How is the block size changed? Extend the devices and, if necessary, the libraries, then select a drive and the properties of the drive.
A block size of 64 kB is recommended for backups of the data of the media agents including Cell Server. However, in order to achieve better performance for the backup of the other servers to be backed up, another logical drive with a high block size can be created in HP Data Protector.
A number of logical drives can be created on a physical drive; the logical drives differ here with regard to the names and selected properties, for example the block size. The device lock name, however, should be identical for the logical drives. Segment Size Increasing the segment size can have a decisive influence on the performance of the data backup and data recovery. A tape or medium is divided into segments: data segments, catalogue segments, and a header segment.
The header information of the tape is stored in the header segment and has the same size as the block size. Data are stored in blocks within the data segments. Information regarding the data segments is stored in the relevant catalogue segment.
The information regarding the data segments is initially stored in the memory of the media agent and subsequently written into the catalogue segment as well as the internal database.
The optimal size of the segment depends on the media type used and the type of data to be backed up. A greater segment size can improve the backup speed.
In the case of backups of a large number of small files, under certain circumstances the segment can be completed sooner. How is the segment size changed? It is recommended to test the performance of the backups for LTO drives in each environment and to increase the size depending on the LTO type used. Tip: A segment size of 8, for LTO drives is a good start value for other performance tests and will also lead to an increase in the backup speed with acceptable performance for recoveries, if current LTO-6 drive technology is used.
Depending on the concurrency, the memory is divided into buffers. Per default, each buffer consists of 8 disk agent buffers that are the same size as the block size. Increasing the number of disk agent buffers does not significantly enhance performance.
However, if the network bandwidth in the case of backups via the network fluctuates greatly, an increase in the number of disk agent buffers can ensure that the media agent always receives sufficient data to operate the drive that is used in the streaming mode.
A reduction in the number of disk agent buffers is necessary if the media agent does not have sufficient shared memory. How are the disk agent buffers changed? Tip: The value 16 for the number of disk agent buffers has proven effective in practice when carrying out network backups. The number of disk agents started per media agent is referred to as disk agent concurrency. The number to be used can be set in the properties of the drive and in the backup specification. A value in the backup specification overwrites the value set on the drive.
HP Data Protector sets a default concurrency when setting up the drives depending on the drive type used.
The default value for LTO drives is 4; if file libraries are used, the default value is 3. The maximum value that can be set is Depending on the backup that is run, the value can be even lower. For example, the value 1 is specified in the backup specification for the backup of a Microsoft Exchange server via the network. However, keep in mind that a value that is set too high can lead to fluctuations in the speed of network backups.
It is recommended to set the value in such a way that at least the drive can be operated in the streaming mode and only set high enough so that the drive that is used reaches saturation. To enhance the backup speed, it is recommended to increase the number of drives for the backup.
If a high concurrency is used during the backup, the restoration can be slower, as the data to be restored are fragmented on the tape. How is the disk agent concurrency changed? Depending on the display, extend the specifications, select a backup specification and select the properties of the job. A value that is set too high can influence the restoration speed, as the data on the tape are fragmented, which causes more rewind time for the tape.
Tip: The test of the concurrency with a null device has proven effective in practice as a suitable test for the optimal setting. Multi-streaming A configuration of multi-streaming can increase the total data throughput rate for the configured mount point and thus contribute to reducing the total duration of the backup. As default, only one disk agent per mount point to be backed up can be started in HP Data Protector.
Multi-streaming multiple data streams enables backing up of a number of parts of a mount point at the same time in that a number of disk agents are started for this mount point. This method is very useful when very large mount points have to be backed up and saturation for the backup device cannot be reached, or when the drive to be used is slower than the disks can deliver their data.
One way around this is to adapt the description of the mount point. In this case, configuration of the backup specification enables individual parts of the mount point to be backed up by a number of disk agents. In this case, configuration of the backup specification enables individual parts of the mount point to be backed up by a number of disk agents on a number of media agents.
How is multi-streaming configured? This operation displays another object. Now, the directories to be backed up can be selected for each mount point. The optimal configuration of multi-streaming depends heavily on the environment involved and should be tested accordingly. The speed of the hard disks on which the data are located and whether saturation of the drive can be reached are two factors to be taken into account for the configuration.
As long as saturation has not yet been reached and the disk agents are still able to deliver sufficient data, an allocation between 2 to 4 disk agents per mount point in combination with the value of the disk agent concurrency and number of deployed drives is a very good setting.
Tip: The configuration is expedient for mount points with a size of more than GB backup volume and if the tape device does not reach saturation. If you plan to use multiple Disk Agents on a share of a local disk, separate the file system in similar slices to parallelize operations correctly.
This allows good restoration speeds using Data Protectors Parallel Restore afterwards. Nice to know, an NDMP overwrite restore on the Celerra takes much more time than a redirected restore into a different directory. Only the catalogue is send to the CS while the NAS head does the whole data movement — directly to tape. Further optimization is possible; it will take time depending on the configuration used. Multiple Disk Agents on the same share are a good option to keep your feed rate high.
My configuration is not ideal to demonstrate this, because three DAs where working on completely different folder structures. This also applies to restores. Everyone who is interested in null device backups on Celerra, try the following command. The backup software will load the cartridge, but no data will be written to tape. This is the only way to verify file system performance during backup time.
As no data is written to media, Data Protector will report such a session as failed. The Get pool and Put pool are good indications if the system performs well or not. You must be logged in to post a comment. Made with by Graphene Themes. Toggle search form Search for:.
0コメント