The…
Up to 99.99% Availability
...
The following setup illustrates a typical bare-metal hardware environment. MediationZone is deployed in at least 1 site/region, with a minimum of 2 nodes.
...
Prerequisites for achieving high availability:
Redundant disk and database storage, that is configured to meet the availability target.
A High Availability Cluster Monitor 3PP. This application monitors process availability and health, and orchestrates an ordered fail-over when an issue is detected:
Migration of the Virtual IP address.
Mounting of storage.
Startup of processes on the stand-by machine.
Up to 99.99% Availability via VM Failover
The…The following setup illustrates a typical Virtual Machine environment. MediationZone is deployed in at least 1 site/region, with a minimum of 2 compute nodes.
...
Prerequisites for achieving high availability:
Redundant disk and database storage, that is configured to meet the availability target.
A 3PP handling Virtual Machine (VM) monitoring and failover to alternate host. For instance VMWare vMotion, VMWare HA, AWS EC2 Auto-scaling, etc. The process monitoring/recovery is achieved by scheduling a script that starts all PICO processes not running on the local VM.
Up to 99.999% Availability Geo-redundant Deployment
...
A geo-redundant setup spans at least 2 sites/regions, with a minimum of 2 compute nodes per site/region. MediationZone can be deployed in an Active-Active or Active-Passive configuration.
...
Active–Active scenarios are typically used for real-time deployments, and consist of two independent sites/installations. Replication requirements:
Filesystem: not required.
System Database: not required.
Audit: not required.
Runtime state: optional, depending on use case. An example of a case where it is required is aggregation/correlation scenarios with database back-end, where the sites act as each others fail-over.
Active–Passive scenarios are typically used for batch applications, and consist of a single Platform instance running. The Passive site is a replica of the Active site, and needs to be started in case the Active site goes down. Replication requirements:
Filesystem: required when runtime data is stored to disk. For example File System aggregation/correlation and Duplicate UDR Detection.
System Database: required.
Audit: optional.
Runtime state: optional, depending on use case.
VM-based High Availability Deployment
...
, Multiple Zones
The deployment illustrated below is a single MediationZone installation spanning across several Availability Zones.
...
Recommendations:
The System Database is handled as a separate cluster (deployed in HA mode), external to the MediationZone deployment.
The Filesystem is mounted to all VMs if file-sharing between VMs is needed (which is common). Also, each VM has local storage embedded in the image.
MediationZone system availability:
In case of a VM failure, the VM is migrated to an alternate host by the virtualization environment. This is required for the ControlZone/Platform, and is use-case dependent for an ExecutionZone.
Workflow availability:
The ControlZone/Platform process orchestrates workflow execution and failover across a set of available ECs.
Real-time collection workflows run as active-active across multiple ExecutionZone nodes. Traffic fail-over between nodes uses:
Protocol-level failover for protocols that support it (Diameter, Radius. etc.)
Load balancer for all other protocols (TCP/IP, HTTP, REST, etc.)
Distributed VMs and ECs over multiple Availability Zones for extra redundancy