Topology
Traditional PBX systems place heavy demands on server resources: CPU, memory, disk I/O, and network bandwidth when handling large volumes of concurrent calls, meetings, queue traffic, IVR interactions, and real-time media processing. As concurrency grows, single-node deployments quickly become a bottleneck, making a clustered architecture essential for scalability, performance, and reliability.
PortSIP PBX is designed from the ground up with a distributed, service-oriented architecture. At the core is a centralized Call Manager, dedicated exclusively to SIP signaling and call control, while specialized functional servers—such as Media, Meeting, Queue (ACD), IVR, Instant Messaging, and DataFlow—operate independently and scale horizontally as a cluster.
This architecture cleanly separates signaling from media and application workloads, enabling efficient resource utilization, predictable performance, and seamless horizontal scaling across large deployments.
Scalability and Capacity
The PortSIP PBX cluster architecture is engineered for carrier-grade environments and can support:
Over 1 million users
Approximately 100,000 concurrent online users (registered or signed in)
Up to 10,000 simultaneous calls
Cluster Architecture Components
The diagram below illustrates the overall PortSIP PBX cluster topology. Each component plays a distinct role in ensuring scalability, performance, and operational resilience.
Call Manager Server
The Call Manager is the central control plane of the system. It is deployed together with the PBX core services, database, and load-balancing logic, and is responsible for:
SIP signaling and call control
Registration and authentication
Routing logic and policy enforcement
Coordination of all backend service nodes
By isolating signaling from media processing, the Call Manager ensures stable call control even under heavy load.
Media Servers
Media Servers handle real-time media processing, including RTP streams, transcoding, recording, and tone generation.
Can be deployed as a single server or as a scalable cluster
Calls are dynamically load-balanced by the Call Manager
Additional Media Servers can be added at any time to increase capacity
This design allows media workloads to scale independently of signaling.
IVR Servers
IVR Servers process interactive voice response logic, prompts, and DTMF input.
Can be deployed standalone or as a clustered service
IVR sessions are load-balanced across available servers
Capacity can be expanded by adding more IVR Servers as required
Queue Servers (ACD)
Queue Servers manage contact center queue logic, including call distribution, agent state handling, and queue metrics.
Support single-server or clustered deployment
Load balancing ensures queues are evenly distributed
Additional servers can be added seamlessly as traffic grows
Meeting Servers
Meeting Servers are responsible for audio and video conferencing workloads.
Meetings are dynamically distributed across the cluster
Capacity scales horizontally by adding more Meeting Servers
Ensures consistent performance for large or concurrent meetings
Instant Messaging (IM) Servers
IM Servers provide real-time messaging and collaboration services.
Typically deployed on a single high-performance server
Can support up to 50,000 concurrent online users
Optimized for high-throughput messaging workloads
DataFlow Servers
DataFlow Servers power analytics, reporting, dashboards, and real-time metrics.
Usually deployed on a dedicated, high-performance server
Designed for high-volume data ingestion and fast query performance
Provides real-time visibility into call activity and system performance
Cluster Architecture with High Availability (HA)
For customers requiring maximum uptime and fault tolerance, PortSIP PBX supports a High Availability (HA) cluster architecture.
Call Manager Servers (HA)
In an HA deployment:
Three Call Manager servers are deployed in an HA configuration
A Virtual IP (VIP) is exposed to client apps, IP phones, and backend services
SIP signaling continues seamlessly even if one Call Manager fails
This ensures continuous service availability and eliminates single points of failure in the control plane.
Detailed design and deployment guidance is available in the PortSIP PBX High Availability Architecture documentation.
Media Servers (HA)
Media Servers are continuously monitored by the Call Manager
If a Media Server fails, active calls are automatically reassigned to available servers
No manual intervention is required for call recovery
IVR Servers (HA)
If an IVR Server becomes unavailable, administrators receive an email alert
The failed server can be disabled through the PBX web portal
IVR traffic is automatically redistributed to remaining servers
Queue Servers (HA)
Queue Servers follow the same high-availability behavior as IVR Servers:
Automatic failure detection
Email alerts to administrators
Manual disablement via the web portal
Automatic reassignment of queue workloads
Meeting Servers (HA)
Meeting Server failures trigger administrator notifications
The failed server can be disabled through the PBX web portal
Active and new meetings are redirected to healthy servers
IM Servers (HA)
IM Server supports up to 50,000 online users per high-performance node
In HA deployments, multiple IM Servers ensure uninterrupted messaging services
Failures do not disrupt user messaging sessions
DataFlow Servers (HA)
DataFlow Server provides analytics and real-time metrics
Typically deployed on powerful dedicated hardware
Designed for continuous availability and high data throughput
Summary
PortSIP PBX’s cluster and HA architecture delivers:
True horizontal scalability
Clear separation of signaling, media, and application workloads
Carrier-grade reliability and fault tolerance
Simple, incremental expansion without service disruption
This architecture makes PortSIP PBX an ideal platform for service providers, enterprises, and UCaaS operators building large-scale, high-performance communication services.
Last updated