Online Tutorials & Training Materials | STechies.com
Register Login

MaxDB High Availability Interview Questions and Answers

|| 0

MaxDB High Availability Interview Questions and Answers
Stechies

 FAQ: MaxDB High Availability

1. Costs and uses for high availability
2. How can I ensure a high availability of the database?
3. Which mechanisms does MaxDB offer to minimize downtimes?
4. Which solution is preferable in which case?
5. How can I speed up backup/recovery?
6. Which solutions are offered for cluster for failover?
7. Which solutions are offered for shadow databases?
8. Which solutions are offered for Hot Standby?
9. Where do I find documentation on the subject of high availability?

Solution

1. Costs and uses for high availability

In general, the more comprehensive the protection against downtime, the more expensive the solution. For this reason, you should first specify your requirements. Relevant are, for example, the maximum unplanned downtime in a certain period and the maximum length of an individual downtime.

For databases, the maximum length for an individual downtime is very important. If this time is shorter than the longest estimated time for the restore, you require additional protection mechanisms.

2. How can I ensure a high availability of the database?

Under UNIX, use raw devices. This is safer than using files in the file system.

For increased safety, use RAID levels for the data volumes. The highest level of safety can be achieved by mirroring the volumes.

Only use RAID levels with striping procedure for the log area of the database if the storage system is big enough. For server systems, the mirroring of the log area is obligatory. We recommend that you carry out a mirroring on the operating system or the hardware. A mirroring of the log area by the database is supported by MaxDB/liveCache. However, it usually leads to a deterioration in performance. For more information on log areas, see Note 869267 and refer to the relevant documentation.

Carry out database structure checks on a regular basis. For more information on this subject, see Note 940420 and refer to the relevant documentation.

3. Which mechanisms does MaxDB offer to minimize downtimes?

Backup/recovery
Implement an appropriate backup/recovery strategy.

Using copied Snapshots or Split Mirror for data backup
MaxDB supports Snapshots and split mirror. For more information on this subject, see Note 371247 and refer to the relevant documentation.

Cluster for Failover
Cluster for Failover makes sure that the database is automatically transferred to a second server, for example, if the database server crashes.

Shadow database (standby database)
A shadow database redoes the transported log files of a master using log recovery. If a master fails, but the current log area is still available, the shadow database can take over the master operation without data loss. The shadow database can also redo the log at a different time and can be switched online starting at an earlier point in time.

Hot Standby
With Hot Standby, MaxDB/liveCache reduces the downtime to a few seconds during the switch to the second server. Prerequisites for Hot Standby are a disk storage system and a library implementation of the disk storage system provider.

With Hot Standby, a database instance works on a second server with a separate data area. In standby, the database reads log entries immediately from the log area of the master and redoes them. After the transfer, the former standby instance continues working with the log of the master.
The data area of the standby instance is automatically initialized using a Flashcopy or a Symclone copy.

A cluster agent monitors the master and automatically switches to standby if the master is not available.

4. Which solution is preferable in which case?

Backup/recovery
Advantages:
- Simple implementation
- No periodic tests required You can check backups online.
- Protects against physical and logical errors.
Disadvantages:
- Downtime for recovery may be long.
- I/O workload during the backup

Snapshots
Snapshots cannot be regarded as data backups unless the data is copied to a second medium.
Advantages:
- Fast recovery of the data
- No synchronization required during Snapshot creation
- Snapshots can be used without a log for system copies
- Snapshots can provide consistent versions of system landscapes
Disadvantages:
- A disk storage system is required
- I/O workload for backup in the storage system
- Database structure check not possible on read only Snapshot
- Does not protect against damaged hard disks

Split Mirror / Snapshot with copy (Flashcopy, Symclone, and others)
Advantages:
- Fast recovery of the data
- No synchronization required during snapshot creation
- Data backup by MaxDB on the mirror possible
- Database structure check on the mirror possible
- Mirrors can provide consistent versions of system landscapes
Disadvantages:
- A disk storage system is required
- The disk storage system must ensure the I/O consistency of all data volumes of a database instance for the time of the split.
- Merge of the mirror into the master
- additional hard disk space for complete mirrors

Cluster for Failover
Advantages:
- Automatic and fast transfer should server problems occur
- Cluster agents available for all platforms
Disadvantages:
- Shared disk approach; data volumes are not protected
- Recovery required if data hard disks are damaged
- I/O workload during backup

Shadow database
Advantages:
- Protects against hard disk and server crashes
- Database structure checks on standby instance possible
- Protects against logical errors (point in time recovery)
- Delay for log recovery determines the maximum downtime
- Implementation possible without further partners
- No performance disadvantages for master operation
Disadvantages:
- No automatic transfer without additional software
- Additional server and hard disk space for standby instance

Hot Standby
Advantages:
- Protects against hard disk and server crashes
- Automatic transfer should the server or the database crash
- Transfer only lasts a few seconds
- Database structure check possible on standby instance
- Standby instance is read only
- No performance disadvantages for master operation
Disadvantages:
- A disk storage system is required
- Implementation must be carried out by hardware partners for HSS library
- Additional server and hard disk space for standby instance

For the protection against disaster recovery (for example, if the computer center breaks down), a shadow database is generally more useful than Hot Standby. If a Hot Standby instance reads from a mirrored log area, the mirror must be mirrored synchronously. As a result, performance may decrease in the master. Downtimes of a few seconds do not have to be ensured for disaster recovery.

5. How can I speed up backup/recovery?

Incremental backups can speed up the recovery of a database considerably.
If you make sure the log area is large, the system does not have to use archived log files for the restore. After a data restore, you can carry out a restart immediately if the starting point for the log recovery is still in the current log.
The bottleneck for backup time or the restore time of the data is usually caused by the access to the backup media. When you use backup servers, ensure a fast connection. Use parallel backup media to speed up the data backup or restore.

6. Which solutions are offered for cluster for failover?

Note 319835 describes the availability of MaxDB/liveCache for Microsoft Cluster.
For all supported UNIX platforms, the operating system partners offer cluster agents for MaxDB/liveCache. Contact your hardware partner.
For Linux, the product LifeKeeper by Steeleye (<http://www.steeleye.com>), among others, supports MaxDB and liveCache.

7. Which solutions are offered for shadow databases?

For shadow databases, Libelle (<http://www.libelle.de>, USA: <http://www.aivant.com />) offer a solution with their product DBShadow.
You have the option of implementing, for example, a script-based shadow database solution. You will find a description in the latest MaxDB documentation. A detailed description is attached to this note (update: April 28, 2008).

8. Which solutions are offered for Hot Standby?

Hot Standby has been implemented for:
- IBM (SVC, ESS, DS8000)
- EMC (Symmetrix)
- HP (StorageWorks XP)

Contact your hardware partner.

9. Where do I find documentation on the subject of high availability?

Note 767598 describes where to find the MaxDB documentation.  .....

 

 


Related Articles