Operax

HA Cluster

Details: Category: SDKs

HA Cluster is a high availability middleware used for monitoring software and restarting it if faults occur, as well as managing failover and switchover of software arranged in 2N (1+1) active/standby instance pairs between two nodes. It is designed for carrier grade performance with short failover times and to work well in realtime-scheduled environments under high loads.

Overview

The tough carrier grade requirements has made HA Cluster a lean design without bloat, necessary to maintain performance and guarantee high quality. This lightweight property makes HA Cluster easy to use and configure, and therefore also an attractive HA platform in less demanding environments. To further enhance the ease-of-use there is a zero-integration option for monitored applications, which is sufficient in many cases. For integrated carrier grade systems there is a HA library that provides an advanced state transition model and per-thread monitoring.

Achieving high availability is not only about using proper software and hardware - the complete system must be configured appropriately to avoid single points of failure, not least the connecting network. HA Cluster provides redundant gateway features built-in which lowers the complexity (and cost) of the connecting network. We also provide our HA expertise on a consultancy basis to help our customers be sure that the total solution software-hardware-network is truly fault-tolerant.

Licensing Options

We provide several licensing options to meet different needs. For personal use and evaluation in the lab the software is free and available for immediate download. For commercial installations we have an additional license, and finally we have an OEM license used when the HA middleware is integrated into a larger product and resold, and in that case we also make the source code available to the customer. We can arrange with source code analysis before buy for the most demanding customers - we understand carrier grade product development and the risks associated with incorporating third-party software for key functions and don't expect anyone to buy the pig in a poke.

Additional Services

We also provide full technical support and professional services for configuration, customization and feature development as well as integration work.

Features

Single node or 2N (1+1) operation
Designed for realtime-scheduled environments with carrier grade requirements
- Maintains responsiveness during high loads, low latency, fast failover times
- Flexible - also works well on Virtual Machines and other non-realtime environments
System model features:
- Restart only (no redundancy) or 1+1 active/standby redundancy model selectable per application
- Add/remove applications in runtime
- Application dependencies (to guarantee start order etc)
- Selectable debug mode (start application in memory debugger for example)
Zero-integration needed for basic monitoring of applications
HA library for enhanced integration with monitored applications
- Complete state transition model, with quiescing state
- Heartbeat monitoring of individual application threads to detect partial lockups
Virtual IP with IP failover available to applications
- Includes safe gratuitous ARP handling - i e packet loss is handled
Virtual gateway with failover
- This means that no HSRP/VRRP configuration of attaching network is required
- Easy-to-configure and lowers cost and complexity of attaching network
Flexible IP management
- Separate management network, dual links etc
- Made for true carrier grade network configurations
- Works well with VLAN configurations
Not only applications are monitored - OS kernel and HA Cluster itself are too
- Lockup of OS kernel or HA Cluster will lead to automatic restart
- Integrated with hardware watchdog (if available)
Monitors key hardware resources
- Gateways, network interfaces, RAM, CPU and storage
- Web services interface for counters, gauges and alarms
Command line interface
- Control individual applications
- Manage system model changes
Event logging