The T-Mesh technology supports multiple TCPWave DDI Management appliances serving the global DDI remotes. The underlying database uses a write-set synchronous replication. The T-Mesh consists of a single floating HA master and multiple HA members. When the floating HA master fails, the next available member automatically assumes the role of a master without any human intervention. The franchise critical DDI transactions do not see any impact when a single floating HA master or a HA member goes down. When a temporary network interruption takes place, the T-Mesh technology auto-recovers, and the DDI management ecosystem is synchronized with a degree of transparency that the end-users see no impact. The T-Mesh cluster maintains a cache to expedite the recovery of a member that has fallen out of sync. The remote DDI appliances are designed to auto-sense the failure of their preferred DDI manager. A HA member failure will automatically prompt an election on the connected remotes to choose their next best DDI manager. The DDI administrator also has the UI ability to swing the DDI remotes from one management node to another without causing any service disruption. The T-Mesh cluster technology is designed to operate with the least amount of configuration changes from the end-users. The T-Mesh cluster self-tunes its configurations every thirty minutes to deliver maximum performance. It is recommended to have three nodes in a T-Mesh. The TCPWave monitoring engine periodically monitors the health of the T-Mesh cluster.
While the T-Mesh technology eliminates the single points of failure in an enterprise, it also provides the DDI administrators with an easy to use user interface to maintain and monitor the cluster. Joining a member or removing a member can be done easily using the web interface. Updates and upgrades, configuration changes, viewing the remote's logs on the web interface, restarting the services, etc. of any DDI remote can be performed from any T-Mesh DDI controller. Internally, the T-Mesh cluster uses a delegate method that hands over the management activity of a given remote from one management node to another. Even though the action is initiated from one management node, the actual node that performs the appropriate action on the remote is the HA node that is directly connected to the remote using the T-Message Secure Tunnel. The transport layer used in the T-Mesh ecosystem is encrypted using the highest degree. The management traffic is encrypted using SSL over a unique TCP port. It is important to note that the nodes in the T-Mesh HA cluster need a proper clock discipline. The clock offset between the nodes of a T-Mesh ecosystem plays a significant factor in the overall stability.
The T-Mesh technology comes with a built-in conflict resolution logic. When two nodes have a dispute, the third node automatically acts as an arbitrator. The dispute resolution takes place in milliseconds. TCPWave recommends that all the nodes on the cluster be on the same network speed, hardware type, patch levels, etc. Since the T-Mesh uses a replicated write-set to guarantee data consistency, a given write operation must be performed on all the nodes. Every DDI operation that uses any of the 1490 plus REST API calls, gets processed within a few milliseconds on the performance-optimized database.