|
|
|
|
The Traverse distributed database and processing architecture allows very high levels of fault tolerance and scalability during deployment. All of the components in the various tiers are horizontally scalable which is essential for expansion and real-time performance reports.
All of the configuration information is stored in the BVE Provisioning Database. On startup, the DGEs connect to the BVE Provisioning Database and download a local copy of their configuration. Any updates made to the BVE Provisioning Database are pushed out in real time to the corresponding DGE.
To handle the case of a DGE physical server going down, you can set up a spare 'hot standby' server in any central location (N+1 redundancy) which has the software installed and configured. In the case of a production DGE going down for an extended period of time due to hardware failure, you can set the name of the DGE in the dge.xml configuration file (see DGE Identity) and start Traverse on the backup server. This backup DGE automatically connects to the BVE Provisioning Database and downloads the configuration of the failed DGE. When the production DGE comes back up, it can be even run in parallel before shutting down the backup DGE. The only caveat is that the performance data collected during this interval will be missing on the production DGE.
If desired, you can have a backup DGE for each of the production DGEs (N+N redundancy) but this is not really needed if the centralized DGE can poll all the data remotely.
If connectivity between the DGE and the BVE database is lost, the DGE continues to poll, aggregate and even generate alarms completely independently. When connectivity to the BVE database is restored, the DGE restarts and downloads a fresh copy of its Provisioning Database.
The BVE database can be replicated on multiple servers for fault tolerance.
The performance database which is local to each DGE can be located on a remote database cluster if needed for fault tolerance also. The JDBC communication between the DGE and the performance database allows such a setup seamlessly just by a few configuration file changes. Contact Kaseya Professional Services for information and pricing for this configuration service.
Lastly, the Web Application and reporting engine also gets all the configuration information from the BVE database server on startup and hence you can have any number of Web Application servers behind a load balancer for fault tolerance as well as distributed report processing.