How does the fault tolerance feature in the Naming Service work?
;; We support a limited form of fault tolerance by implementing failover feature in our Naming Service. The approach taken for providing failover capability is using a Master/Slave model. This requires the user to keep two Naming Servers running at the same time, with one of them in standby mode. Both the Master and Slave Naming Servers will support the same underlying data in a persistent backing store. And for this scheme to works, the caching facility for both Naming Servers have to be turned off. This would force both Naming Servers to deal directly with the backing store, ensuring that its data remains constant. Failover support doesn’t make sense for a non-persistent backing store since all the data is local to the Naming Server. In this model, only the Master Naming Server is in active mode while the Slave Naming Server plays a secondary role. ;;; When both the Naming Servers are active, the Master is always the preferred choice for the clients that are using Naming Service. In
We support two forms of fault tolerance: by implementing failover feature in our Naming Service. The approach taken for providing failover capability is using a Master/Slave model. This requires the user to keep two Naming Servers running at the same time, with one of them in standby mode. Both the Master and Slave Naming Servers will support the same underlying data in a persistent backing store. And for this scheme to work, the caching facility for both Naming Servers have to be turned off. This would force both Naming Servers to deal directly with the backing store, ensuring that its data remains constant. Failover support doesn’t make sense for a non-persistent backing store since all the data is local to the Naming Server. In this model, only the Master Naming Server is in active mode while the Slave Naming Server plays a secondary role. When both the Naming Servers are active, the Master is always the preferred choice for the clients that are using Naming Service. In the event th
We support a limited form of fault tolerance by implementing failover feature in our Naming Service. The approach taken for providing failover capability is using a Master/Slave model. This requires the user to keep two Naming Servers running at the same time, with one of them in standby mode. Both the Master and Slave Naming Servers will support the same underlying data in a persistent backing store. And for this scheme to works, the caching facility for both Naming Servers have to be turned off. This would force both Naming Servers to deal directly with the backing store, ensuring that its data remains constant. Failover support doesn’t make sense for a non-persistent backing store since all the data is local to the Naming Server. In this model, only the Master Naming Server is in active mode while the Slave Naming Server plays a secondary role. When both the Naming Servers are active, the Master is always the preferred choice for the clients that are using Naming Service. In the eve
