The Subtleties: 2110 vs. NMOS vs. Orchestration

The Stage

Think of it like a professional orchestra:

  ST 2110 (The Musicians): This is the raw "essence" (Video, Audio, Metadata) moving as independent RTP streams. It is the transport standard, but it has no "brain"—it doesn't know where to go.

NMOS (The Sheet Music): Developed by AMWA, NMOS (Networked Media Open Specifications) is the open language that allows devices to talk to each other.

• IS-04: "I am here" (Discovery & Registration).
• IS-05: "Send your video to this IP" (Connection Management).

Orchestration (The Conductor): This is the software layer that tells the NMOS "Sheet Music" what to do and ensures the network "Stage" (Cisco/Arista) is ready for the performance.

Or you can look at it from a "what happens when I  push the button?!"

Below is the control stack in a nutshell

In the ST 2110 ecosystem, the "chain" touches almost every layer of the OSI model.

1. The Control & Orchestration Plane (Top Layers)
Layer 7 (Application): the "Brain" or the "Island" of the ship
Orchestration Software: Brands like Riedel, GV Orbit, or Ross Ultricore run as applications here. They manage the "logic" of the facility.

NMOS (IS-04, IS-05, IS-08): These are HTTP/RESTful APIs. When a browser or control panel says "Route Camera 1," it is a Layer 7 action.

Layer 6 (Presentation):
SDP (Session Description Protocol): This defines the "flavor" of the stream (e.g., "I am 10-bit 4:2:2 1080i video"). It translates technical metadata so the receiver knows how to "present" the data.

2. The Network & Session Plane (Middle Layers)
This is the "Nervous System." It manages the connections and the paths the data takes.

Layer 4 (Transport):
RTP (Real-time Transport Protocol): The "envelope" for ST 2110. Every video/audio packet is wrapped in an RTP header to keep track of sequence and timing.

UDP (User Datagram Protocol): Media is sent via UDP because we can’t wait for "re-transmissions" (TCP) in live TV.

Layer 3 (Network):
IP (Internet Protocol): Every device has an IP address. This is the global "postal system."

IGMP (Internet Group Management Protocol): This is the crucial link. It is the "request" from a monitor to a switch saying, "I want to join this multicast group." It bridges the Control Plane to the Network Fabric.

PTP (Precision Time Protocol / IEEE 1588): While it relies on Layer 2 for some hardware-stamping, it operates as a Layer 3 protocol to synchronize the entire facility to a single Grandmaster clock.

3. The Media & Physical Plane (Bottom Layers)
This is the "Hull" and the "Engine Room." It is where the physical reality of the packets meets the wire.

Layer 2 (Data Link):
Ethernet (MAC Addresses): The switches (Cisco/Arista) use MAC addresses to move frames between ports. This is where VLANs and high-speed switching happen.

Layer 1 (Physical):
Fiber / SFP+ / QSFP: The actual laser light or electrical pulses moving at 10Gbps, 25Gbps, or 100Gbps.

The Riedel / Cisco Relationship: Is it a Web Server?

  First let's look at a common division of "labor between the orchestration, NMOS, and switching layers.

The relationship between Riedel and Cisco in an ST 2110 environment is a classic example of the "Separation of Planes." To understand this, we have to look at the Control Plane (the brain) and the Data Plane (the muscle).

The "Web Server" Misconception
When engineers interact with Riedel orchestration software—such as Explorer or Director—they are often using a web-based GUI. This creates the illusion that the "Web Server" is the thing routing the video.

In reality, the Riedel Orchestrator acts as a centralized logic engine. It hosts a web interface for user interaction, but its primary job is to serve as an API Gateway between human intent and network execution. It never touches the actual video essence; it only manages the "permission" for that video to move.

The Process: From Browser to Cisco Router
The journey from a user click to a multicast flow involves several distinct steps:

• 1. The HTTP/Web Request
  The process begins in a web browser. When an operator selects a source and a destination on the Riedel interface, the browser sends an API command (typically via REST or WebSocket) to the Riedel Orchestrator.
• 2. Pathfinding and Logic
  The Riedel "Brain" consults its internal map of the facility. It checks:Availability: Is the destination already receiving another stream?Bandwidth: Is there enough capacity on the specific network links to support the requested format (e.g., 12G-SDI vs. HD)?
• 3. The SDN / API Handshake
  Instead of the user logging into a Cisco switch via a Command Line Interface (CLI), the Riedel Orchestrator talks to the Cisco Nexus Dashboard or DCNM (Data Center Network Manager).The orchestrator says: "I have authorized a 12Gbps flow from Port A to Port B. Please clear the lane."
• 4. Cisco Multicast Execution
  The Cisco Nexus switches receive this instruction and update their IGMP/PIM tables. The switch then handles the high-speed packet duplication.The Riedel software ensures the "Logic" is sound.The Cisco hardware ensures the "Physics" (the packets) are delivered.

Summary of Roles

By separating these layers, the system remains Deterministic. Even if the Riedel "Web Server" were to be disconnected, the existing video flows on the Cisco fabric would continue uninterrupted, as the switch already has its instructions.

1. The User clicks a button on a Riedel panel to route Camera 1 to Monitor 1.
2. The Riedel Orchestrator calculates the path.
3. The Orchestrator talks to Cisco (via API/SDN): It tells the Cisco Nexus switch, "I need 12Gbps of bandwidth from Port 1 to Port 10. Please clear the lane."
4. Cisco handles the IGMP/PIM (Multicast): The switch does the heavy lifting of moving the packets, while the Riedel software acts as the brain overseeing the logic.

Is GV AMPP an Orchestration System?

Yes, but it's a "Super-Orchestrator."

As you can see from the Grass Valley AMPP page you’re viewing, AMPP is a Cloud-Native Live Production Platform.

While a system like Riedel might focus on routing hardware signals in a building, AMPP orchestrates microservices. It spins up a virtual switcher, a virtual multiviewer, and virtual audio mixers in the cloud or on-prem. It handles the essence flows (the 2110/SRT/NDI streams) between these virtual bits of software. It is orchestration at the application level, not just the network level.

NEP’s TFC: The "Manager of Managers"

NEP’s TFC (Total Facility Control) is what we call a Control System or User Interface Layer.

NEP builds massive mobile units and broadcast centers. They don't want their operators to have to learn the deep technical menus of Grass Valley, Riedel, and Cisco separately.

• TFC sits on top.
• It talks to GV AMPP for cloud workflows.
• It talks to Riedel for comms and gateway routing.
• It talks to Cisco/Arista for the network backbone.

Ross Video: The Hyperconverged Philosophy

While Grass Valley separates the "Ground" and "Cloud," Ross Video focuses on Hyperconvergence—consolidating the router, switcher, and multiviewer into unified hardware.

The Ross Video Ultrix hyperconverged platform is designed to consolidate a massive amount of traditional broadcast hardware into a single, software-defined chassis.

1. The Unified Backplane (The Core)
At the center of the platform is the ST 2110 / SDI / NDI / Dante Unified Backplane. Unlike traditional routers that only move signals, this backplane allows different signal types to coexist. You can ingest 12G SDI cameras alongside ST 2110 essence flows, and the system treats them as native internal sources.

2. Multi-Format I/O (The Blades)
The graphic shows the modular "blades" or cards that slide into the frame:

• Ultrix-IP: These cards handle the heavy lifting of ST 2110 ingest and egress, acting as a high-density gateway.
• Hybrid Connectivity: It supports remote feeds via Ross Catena, allowing cloud-based microservices to bridge into the physical hardware.

3. Integrated Software Microservices
The "Hyperconverged Efficiency" comes from licensing software to perform tasks that used to require separate racks of gear:

• Ultriscape: This is an integrated multiviewer. It "spits out" multiviewer heads directly from the router without needing external cabling or additional latency.
• Ultrimix: An advanced audio processing engine that can shuffle audio between SDI, 2110, MADI, and Dante streams seamlessly.
• Ultra-Signal-Processing: This handles critical timing tasks like Frame Sync and HDR Conversion internally on the backplane.

4. The Hierarchy of Control
The diagram above breaks down the management stack:

• Application: Ross Catena manages the high-level workflow.
• Control: Ultricore BCS/IP acts as the "Brain," handling NMOS registration and pathfinding.
• Data Plane: The Ultrix Frame itself, where the actual data switching occurs.
• Physical: The underlying Cabling & Network Fabric.

Why "Hyperconverged"?
This approach is about Efficiency. By keeping the routing, switching (Carbonite), and multiviewing (Ultriscape) inside one box, you eliminate thousands of cables and multiple points of failure. It provides "Deterministic Distribution," meaning the signals move with predictable, ultra-low latency because they never have to traverse an external network to be processed.

• The Brain: handles registration and pathfinding across the network fabric.
• The Hub: The router acts as a software-defined engine. It processes without the signals ever leaving the box.
• The Benefit: This "box-in-a-rack" approach reduces the complexity of managing thousands of external multicast IP addresses, making it ideal for OB trucks and fixed studios.

The "Switzerland" Vendors: Specialist Orchestrators

In a "Best of Breed" facility, you may use a third-party "Manager of Managers" to talk to everyone else:

• EVS Cerebrum: Known for its customizable UI, it can control a Ross switcher, GV router, and a Calrec console from one interface.
• Lawo HOME: A cloud-native platform specialized in "quarantining" and securing new devices as they join the network.

Orchestration in 2026: AI & Federation

  The conversation at NAB 2026 has shifted toward automation and global connectivity:

The AMPP Federation allows these two distinct layers to "handshake" securely. It means a ground-based system (like an NEP mobile unit running TFC) can securely link its local signal orchestration with a broadcaster’s cloud-based AMPP instances.

  The Result: You can move signals from a physical camera on the ground into a virtual switcher in the cloud and back again without manual "translation" or jumping between different control interfaces.

The industry is shifting from simple routing to deep device control:

• The Signal Level (Traditional):
  This is the baseline for most 2110 systems. It uses NMOS IS-05 to manage the "flow." You are essentially telling a receiver to join a multicast group. You have moved the video, but you haven't "controlled" the device.
• The Device Level (The 2026 Standard):
  This is where NMOS IS-12 comes in. It allows the orchestrator to go beyond routing and actually "reach inside" the hardware.

In practice: Instead of just routing a camera signal, a "Device Level" orchestrator allows you to change the camera's gain, shutter speed, or tally status from the same unified interface (like EVS Cerebrum or GV Orbit) without needing the manufacturer's specific remote control panel.

This highlights that the "Silo" approach—where you need a different app for every brand of gear—is crumbling in favor of this deeper, device-level orchestration.

How did the ecosystem get this way?

  It evolved through three distinct phases:

The Result: We now have a "Best of Breed" ecosystem. You use Cisco for the "Big Pipe," Riedel or Imagine for the "Signal Management," and GV AMPP or TFC for the "Creative Interface." They all talk to each other over the network using APIs and NMOS.

AMPP and ST 2110: How They Complement Each Other

Think of ST 2110 as the "Physical Road" and AMPP as the "Self-Driving Delivery Service" that runs on it.

ST 2110 is the Standard for Transport: It defines how uncompressed "essence" (video, audio, and data) is chopped into packets to move across a high-speed network. It is "dumb" transport—it moves the data but doesn't know why it’s moving it.

AMPP is the Operating System: It is a software-defined platform that handles the processing (switching, mixing, graphics). AMPP can take in ST 2110 streams from your cameras, process them in a virtual switcher, and then output an ST 2110 stream back to your monitors.

The "Handshake": Grass Valley uses AMPP Edge as a bridge. An AMPP Edge node sits on your local network, ingesting ST 2110 streams from your physical hardware and "on-ramping" them into the AMPP software environment. This allows you to mix local uncompressed 2110 sources with remote SRT or NDI feeds seamlessly.

When Does One Take Precedence?

  While they work together, the "Priority" shifts based on your Latency requirements and Location.

Scenario A: ST 2110 Takes Precedence (The Hardware-First Build)

• When: You are building a permanent, high-end broadcast center or a Tier-1 OB Truck.
• Why: You need zero-latency and uncompressed quality.
• The Workflow: You prioritize the network infrastructure (Cisco/Arista switches) and physical hardware (dedicated Switcher frames, Multiviewers). You use NMOS to manage these connections. AMPP might only be used as a "backup" or for a secondary "social media" cut of the show.

Scenario B: AMPP Takes Precedence (The Software-First Build)

• When: You are doing a remote production, a "pop-up" event, or a show that needs to scale up and down quickly (e.g., a 3-day music festival).
• Why: You don't want to ship 20 racks of gear. You want to pay for what you use (Pay-As-You-Go).
• The Workflow: AMPP is the "brain." You might use local ST 2110 for your cameras at the venue, but all the actual work (the switching and audio mixing) happens in the AMPP cloud or a centralized data center. In this case, 2110 is just the "feeder" into the AMPP ecosystem.

  Summary: Which to Choose?

NMOS IS-04 Comparison: GV Orbit vs. Ross Ultricore

GV Orbit is generally more "Network-Agnostic": It is built to be the brain for a massive, multi-vendor network where devices are constantly being added and moved.

Ross Ultricore is "Fabric-Optimized": It excels when it is managing the Ross Ultrix backplane, making the registration of internal modules (like the Carbonite switcher) instantaneous and invisible to the user.

In 2026, the best orchestration isn't the one with the most features; it's the one that makes the underlying IT complexity invisible so folks can focus back on the being creatative.

The Bottom Line: You use ST 2110 when you need to "own the wire" and want the absolute fastest, highest-quality path between two points in a building. You use AMPP when the "wire" doesn't matter as much as the flexibility of where and how you produce the content.