Streaming server deployment

If your deployment requires more than one server, you realize the following benefits from using a streaming server deployment rather than a synchronization server deployment (which is obsolete): greater availability of data, cost savings, and elimination of merging problems when consolidating data.

During discovery, data flows in parallel (or streams) from multiple discovery servers to the primary storage server, where the data is processed and stored in the database. Only the primary storage server has a database. Discovery servers are used only for running sensors and therefore do not have a database.

This type of deployment therefore provides the following benefits:
Greater availability of data
In a streaming server deployment, data is available as soon as it is discovered.

In contrast, in a synchronization server deployment, the data for a specific domain is unavailable until the synchronization occurs, and data is also unavailable during the synchronization.

Cost savings
A streaming server deployment requires less hardware and resources.
Elimination of merging problems when consolidating data
In a streaming server deployment, data streams directly to the primary storage server, which prevents the following issues that can occur in a synchronization server deployment:
  • Complicated merging scenarios
  • Problems that occur if domains overlap

The primary storage server is the coordinator of the storage server pool, which is a cluster of storage servers. Each secondary storage server registers with the primary storage server, and each discovery server is notified when a secondary storage server is added to, or removed from, the storage server pool.

If a discovery server has problems when trying to contact a specific secondary storage server, it tries to contact a different secondary storage server, and it continues this process until it succeeds. When a new storage server joins the storage server pool, the discovery server is notified by the primary storage server. The discovery server then contacts the new storage server.

In a streaming server deployment, you use the Data Management Portal (web-based user interface) to view discovery, topology, reporting, and analytical information. You use the Discovery Management Console (client user interface) to perform discovery-related activities, such as the following activities:
  • Starting a discovery
  • Showing the progress of a discovery
  • Managing discovery scopes
  • Managing discovery profiles
  • Managing access list information
discovery server
A TADDM server that runs sensors in a streaming server deployment but does not have its own database.
storage server
A TADDM server that processes discovery data that is received from the discovery servers and stores it in the TADDM database. The primary storage server both coordinates the discovery servers and all other storage servers and serves as a storage server. All storage servers that are not the primary are called secondary storage servers.
streaming server deployment
A TADDM deployment with a primary storage server and at least one discovery server. This type of deployment can also include one or more optional secondary storage servers. The primary storage server and secondary storage servers share a database. The discovery servers have no database.

In this type of deployment, discovery data flows in parallel from multiple discovery servers to the TADDM database.

In a streaming server deployment, the following TADDM server property must be set to one of the following values:
  • com.collation.taddm.mode=DiscoveryServer
  • com.collation.taddm.mode=StorageServer
For all servers except for the primary storage server, the following properties (for the host name and port number of the primary storage server) must also be set:
  • com.collation.PrimaryStorageServer.host
  • com.collation.PrimaryStorageServer.port

If the com.collation.taddm.mode property is set, the com.collation.cmdbmode property must not be set or must be commented out.

Figure 1 is a simple illustration of a streaming server deployment that shows the information flow from the discovery servers to the storage server and its database. Figure 2 provides more detail. It shows the information flow from the discovery servers, each with a Discovery Management Console, to the storage server pool, which includes multiple storage servers and one primary storage server. Each storage server has a Data Management Portal, and all storage servers share one database.

In the Data Management Portal, a storage server is shown as part of the Storage Pool Member list under either of the following conditions:
  • The storage server is not building topologies
  • The value of the TADDM server property com.collation.AlwaysBusyStorageServer in the collation.properties file is false.
If the value of the com.collation.AlwaysBusyStorageServer property is true, the storage server is not shown as part of the Storage Pool Member list because it is dedicated to graphical user interface and integration services.
Figure 1. Streaming server deployment
Streaming server deployment showing the information flow from the discovery servers to the storage server and its database
Figure 2. Streaming server deployment in more detail
Streaming server deployment showing the information flow from the discovery servers, each with a Discovery Management Console, to the storage server pool, which includes multiple storage servers and one primary storage server. Each storage server has a Data Management Portal, and all storage servers share one database.