Trillium | Distributed Fault-Tolerant/High-Availability (DFT/HA) Core
Description
The Distributed Fault-Tolerant/High-Availability (DFT/HA) core software functionality allows the creation and management of:
- Distributed Fault-Tolerant applications
- Pure Distributed applications
- Pure Fault-Tolerant applications
For Distributed Fault-Tolerant applications, illustrated in Figure 1, the DFT/HA Core software distributes the protocol load onto available physical processors; dynamically re-distributes the load on processor failure and new processor introduction; retains active calls and recovers from processor failure; and allows maintenance operations to be performed without bringing the system down.
Figure 1: Distributed Application
For Pure Distributed applications, the DFT/HA Core software distributes the protocol load on available physical processors and dynamically re-distributes the load on processor failure and new processor introduction.
For Pure Fault-Tolerant applications, illustrated in Figure 2, the DFT/HA Core software can be used to build an active/standby redundant system. The core software manages system operations, retains active calls on processor failure, and allows maintenance operations to be performed without bringing the system down.
Figure 2: Pure Fault-Tolerant Application, SS7 stack
DFT/HA Core software enables telecom equipment manufacturers of soft switches, signaling gateways, signal transfer points, service control points, media gateway controllers or other products to:
- Accelerate time to market
- Reduce development costs
- Reduce project risk of internally developed DFT/HA applications
Product deliverables consist of C source software, documentation, training, a warranty and technical support.
Features + Benefits
- Multiple system configuration support. The DFT architecture supports all standard DFT configurations in which a DFT protocol layer may reside. For applications that do not fall into one of the pre-defined system configurations, custom configurations may be created.
- Automatic system initialization. The core software is capable of starting active and standby copies of protocol layers on physical processors based on the selected system configuration. The protocol layer load is equally distributed on all available processors. If a node fails and is removed from the system or a new node is introduced into the system, the protocol load is redistributed on available processors. This load re-distribution is performed without interrupting ongoing protocol operations and is completely transparent to the system user. A protocol layer's load is re-distributed to maintain its preset system configuration. The same operation may be initiated when a node is lost due to system failure.
- Protocol layer specific configurations. The layout of one protocol layer in the system is completely de-coupled from the layout of other protocol layers in the system. It is possible to set up one protocol layer in a pure fault-tolerant configuration while having another protocol layer of the stack in an asymmetric distributed fault-tolerant configuration.
- Multiple choices of load distribution criteria. Each protocol layer may define one or more criteria upon which the load distribution is performed.
- Complete recovery during failure. This feature is available in both pure fault-tolerant and distributed fault-tolerant systems. When a failure occurs, failed protocol layers are able to completely recover stable state information. All protocol resources present in a stable state during the failure are maintained on the standby.
- Application restart on processor loss. This feature is applicable to pure distributed systems. If a processor in a pure distributed system fails, applications on the failed processor may be restarted on available processors to provide service for subsequent user traffic.
- Survive up to n-1 faults. DFT protocol layers may survive up to n-1 faults without loss of service where n is the number of processors over which the protocol layer was distributed. With the lost application restart feature enabled, a distributed protocol layer may continue to provide full service until the last processor in the system fails.
- User defined system operations. Advanced distributed system operations such as dynamic load balancing may be implemented using basic services provided by the core software.
- Graceful node shutdown. The system manager provides an operation to gracefully shutdown a node and an option to redistribute the protocol load onto remaining processors in the system. The load redistribution is completely transparent to the system users.
- Maintenance operations. The system manager provides an operation to swap the states of an active and standby node. This functionality may be used to perform maintenance operations on the system without shutting it down. These operations are completely transparent to the system users and will not interrupt service provided by the system.
- Supports software upgrades using the Rolling Upgrade feature. This feature allows upgrading to a new version of the software without taking the system offline or interrupting service.
- Alarms for failure detection. Each core component, protocol layer and protocol layer component generates a set of alarms when the component encounters an unexpected condition. These alarms may be used by the stack manager to passively detect faults in interacting system components.
- Statistical information. Each core component, protocol layer and protocol layer component maintains statistics information about ongoing operations. This information may be collected by the stack manager and used to determine system performance and processor loading in real-time. This information may be used to perform dynamic load balancing by the stack manager.
- Conforms to Trillium Advanced Portability Architecture (TAPA)
- Benefits of licensing Trillium software from Continuous Computing
Product Interworking
Trillium DFT/HA Core software works directly with:
- Fault-Tolerant/High-Availability (FT/HA) Protocol Specific Function (PSF) to provide fault-tolerant, High Availability functionality for specific protocol layers.
- Fault-Tolerant/High-Availability (FT/HA) Protocol Specific Function (PSF) and Load Distribution Function (LDF) to provide distributed and/or fault-tolerant, High Availability functionality for specific protocol layers.
DFT/HA Core software directly supports Trillium software products that must be distributed (pure distribution) and/or distributed and protected from failure (distributed fault-tolerant).
Continuous Computing also offers Professional Services to integrate Trillium software into customers' products.
Conformance
Trillium DFT/HA Core software conforms to the following standard:
- Trillium Advanced Portability Architecture (TAPA)
