Difference between revisions of "Additional Deployment Options"
From AgileApps Support Wiki
imported>Aeric |
imported>Aeric |
||
(3 intermediate revisions by the same user not shown) | |||
Line 10: | Line 10: | ||
:* Add [http://www.scaledb.com/pdfs/ArchitecturePrimer.pdf Database Clustering].<br>With a database cluster, the application server sees a single database, but inside the cluster many database servers are at work. | :* Add [http://www.scaledb.com/pdfs/ArchitecturePrimer.pdf Database Clustering].<br>With a database cluster, the application server sees a single database, but inside the cluster many database servers are at work. | ||
{{Note|All of those options are in effect in the public cloud installation of the {{EnterpriseBrand}}, with the exception of database clustering. (So far, the performance profile has not required it, despite having thousands of users on the system.)}} | {{Note|<br>All of those options are in effect in the public cloud installation of the {{EnterpriseBrand}}, with the exception of database clustering. (So far, the performance profile has not required it, despite having thousands of users on the system.)}} | ||
'''Additional DB Clustering Notes:''' | |||
:* Third party database clustering systems work well. | |||
:* The {{EnterpriseBrand}} currently supports MySQL version {{MySqlVersion}}. But MySQL native clustering becomes stable only in version 5.6. | |||
:* Whichever system is used for DB clustering, it should support ''multi-master replication''. Otherwise, separate replication nodes are needed for every system in the cluster, and failover procedures become tricky. | |||
:* The Quartz scheduler has yet to be tested for scheduled reports and backups when DB clustering is in effect. |
Latest revision as of 23:27, 1 July 2015
To further improve the performance of the system, data safety and data security, there are some additional things you can do:
- Send database backups to a distant, remote server.
Keep the replicated database on premise, for instant failover when needed. But for disaster recovery, keep the backup database 1000 miles away, or more.
- Send database backups to a distant, remote server.
- Allow only the backup server to connect to the backup port.
In addition to requiring the username and password to connect to the backup port, allow connections only from a specific IP address.
- Allow only the backup server to connect to the backup port.
- Add a Dynamic Site Accelerator--in effect, a giant cache in the cloud
For example, use the Akamai accelerator.
- Add a Dynamic Site Accelerator--in effect, a giant cache in the cloud
- Add Database Clustering.
With a database cluster, the application server sees a single database, but inside the cluster many database servers are at work.
- Add Database Clustering.
Additional DB Clustering Notes:
- Third party database clustering systems work well.
- The AgileApps Cloud platform currently supports MySQL version 5.5. But MySQL native clustering becomes stable only in version 5.6.
- Whichever system is used for DB clustering, it should support multi-master replication. Otherwise, separate replication nodes are needed for every system in the cluster, and failover procedures become tricky.
- The Quartz scheduler has yet to be tested for scheduled reports and backups when DB clustering is in effect.