Direct Connection
Cameras connect over fiber or high-availability broadband straight to the private cloud, with no servers on-premises. The simplest model where the link supports it.
Three Ways to Connect a Camera
A city is not one network. Some sites have fiber, some have a congested radio link, and some have neither reliably. IREX supports three connectivity models so each site gets the one it can actually sustain.
The Choice
Bandwidth Decides, Not the Contract
All three models provide centralized management, high availability, access logs, and web-based access from any device. What changes is where the analytics run and how much traffic crosses the link.
The commercial model is independent of the connectivity model. A self-hosted subscription, a perpetual license, or IREX-hosted SaaS can each use any mix of the three, and a single deployment routinely uses all three across different sites.
The Peru deployment is the cautionary tale we quote internally: edge servers that were in the requirements but not in the go-live scope. Infrastructure that is not priced is not infrastructure you avoid, so the site survey in the design phase exists to settle this before contract.
The Three Models
Pick One per Site
High-Availability Edge
Redundant local edge servers run analytics and storage close to the cameras, over wired or wireless networks. For bandwidth-constrained or unreliable links where the site cannot afford to go dark.
Cost-Effective Edge
A single edge server handles local analytics and storage, optimized for smaller sites with limited connectivity where redundancy is not justified.
Why Edge
What an Edge Server Changes
- Only events travel to the cloud, not raw video, which collapses the bandwidth requirement.
- The site keeps recording offline when the network drops, and synchronizes events automatically once it recovers.
- Analytics latency stays local, which matters where the alert has to reach an operator on the site itself.
- Central management, search, and the audit trail still cover edge sites: it is one system, not an island.
- Redundant edge servers remove the single point of failure at sites that cannot tolerate one.
- Edge hardware is scoped explicitly in the design phase and priced, never assumed.
Facts
- Common to All Three
- Centralized management, high availability, access logs, and web access from any device.
- Site Survey
- Network discovery plus a physical and environmental assessment fixes the model per site before the statement of work is signed.
- Timeline
- Six to twelve months from first qualified conversation to steady-state operations, with the procurement route setting the pace.
FAQ
How do we know which model a site needs?
The site survey in the system-design phase settles it: network discovery, plus a physical and environmental assessment and an integration inventory. IREX then produces the target architecture, the hardware and software bill of materials, and the capacity model.
Can one deployment mix models?
Yes, and most do. The connectivity model is chosen per site, and the commercial model is independent of it.
What happens at an edge site when the link drops?
The edge server keeps recording offline and synchronizes its events automatically once the network recovers, so the archive has no hole.
Who supplies the edge hardware?
It is scoped and priced in the system-design phase. It can come through IREX, through a local partner, or from the customer directly, but it is never left out of scope: unpriced infrastructure has cost a deployment before.
Start with a Site Survey
Bring the network as it is, not as the documentation says it should be. That is what makes the capacity model worth anything.