Data centers operate under a different set of rules than most facilities when it comes to physical security. The stakes are higher, the compliance requirements are more demanding, and the tolerance for downtime including security system downtime is essentially zero. When organizations in this space start evaluating their video surveillance infrastructure, they often arrive at one of two conclusions: they assume they need to keep everything on premises because the cloud feels too risky, or they assume a full cloud migration is the modern answer. In most cases, both conclusions are wrong.
The right model for data centers is hybrid, but hybrid done deliberately, not hybrid by accident. There’s a meaningful difference between a thoughtfully architected system that combines on-premises control with cloud-enabled flexibility and a patchwork approach that ends up creating management complexity without delivering the benefits of either model. In this first part, we’ll look at why both extremes fall short and what a well-designed hybrid model actually looks like.
Why Pure Cloud Isn’t the Right Answer for Most Data Centers
The appeal of a fully cloud-hosted video security system is understandable. Reduced on-site infrastructure, remote accessibility, simplified management — these are real benefits, and they matter. But data centers face constraints that make a pure cloud model difficult to justify.
The first is bandwidth. A modern data center may have hundreds of cameras deployed across server halls, loading docks, cage areas, perimeter access points, and executive corridors. Streaming all of that footage continuously to a cloud platform consumes significant network resources and that’s before accounting for how to manage it intelligently.
Most modern IP cameras support multiple simultaneous streams at different resolutions and frame rates, and a well-configured VMS takes advantage of this. Under normal, non-event conditions, CompleteView monitors and records a camera’s lower-resolution, lower-frame-rate stream which consumes a fraction of the bandwidth and storage of a full-quality feed. When an alarm or event fires, for example a motion trigger, a door contact, an analytics alert, CompleteView automatically switches to that same camera’s high-resolution, high-frame-rate stream to capture the detail that investigation and compliance require. The camera never changes; only which stream the system is paying attention to does.
At 200 to 500 cameras, that distinction matters enormously. A deployment that records every camera at full resolution and full frame rate around the clock will consume multiples of the storage and bandwidth of one configured with intelligent multi-streaming. In an environment where network performance is a core business metric, routing undifferentiated high-bitrate video to the cloud is not a tradeoff most operators are willing to make. With the right on-premises VMS configuration, they don’t have to.
CompleteView adds a second layer of bandwidth efficiency through Dynamic Resolution Scaling (DRS), a technology proprietary to Salient. When a client workstation requests video from a recording server, it simultaneously communicates the current size of the display window. The recording server then delivers only the pixel data needed to fill that window, nothing more. An operator monitoring a small thumbnail view of a camera gets thumbnail-appropriate data. If they pull that feed up full screen, the server scales accordingly. The practical effect is that the system never transmits more data than the screen can display, which across a large multi-monitor security operation adds up to significant bandwidth savings without any loss in perceived video quality.
The second is data sovereignty and compliance. Data centers are frequently subject to SOC 2, ISO 27001, PCI-DSS, and in some cases HIPAA requirements, depending on the tenants they serve. Many of these frameworks require demonstrable control over where data is stored, who can access it, and how access is logged. When video data leaves the building and resides on a third-party cloud platform, that control becomes harder to demonstrate and harder to audit. Compliance teams and security officers at data centers tend to be skeptical of any architecture that moves sensitive operational data off-premises without a compelling reason.
The third is latency. Access control and video surveillance in a data center need to operate in near real-time. If a door event triggers a camera to pull up video and an operator needs to make an access decision, delays introduced by cloud routing can erode the effectiveness of the system. Local processing ensures that the critical, time-sensitive functions happen where they should, on-site.
Why Pure On-Premises Falls Short Too
If cloud introduces risk, the natural instinct for many data center security managers is to keep everything local and avoid the complexity. That’s a defensible position, until the operational realities of running a multi-site or enterprise-scale video deployment start to surface.
Managing an on-premises-only system across multiple data center locations means maintaining separate server infrastructure at each site, managing individual system updates, and having no centralized visibility unless someone has built a custom integration layer. When a camera goes offline at 2 a.m. at a remote facility, someone has to notice and in an on-premises-only model, that usually means waiting for an alert that may or may not have been configured correctly, or relying on staff to catch it during a manual check.
Remote access in an on premises model typically requires a VPN. That works, but it introduces its own management burden VPN credentials to maintain, firewall rules to manage, and IT involvement every time access parameters change. In practice, many organizations end up with security staff who can’t easily access the system from outside the facility, which defeats the purpose of having remote monitoring capability at all.
On premises also tends to create silos. Each location operates as its own island, and getting a unified view across sites requires deliberate architectural work that not every deployment receives.
What a Well-Designed Hybrid Model Looks Like for Data Centers
A hybrid video security model for data centers keeps the right things local and extends the right things to the cloud. The distinction matters, because hybrid isn’t just about splitting storage between two environments, it’s about allocating function based on where each function performs best.
Local control stays on premises. Video recording, storage, and real-time processing should happen at the facility. This preserves the low latency required for integrated access control decisions, keeps recorded footage under the operator’s direct control, and ensures compliance with data residency requirements. The on-premises system operates independently if the cloud connection is interrupted, the system continues to function without degradation.
Remote access and management move to the cloud. This is where cloud delivers genuine value for data centers without introducing risk. Operators can monitor system health, view live or recorded footage, manage user permissions, and receive alerts from anywhere, without a VPN and without routing video traffic through the cloud during normal operations. A well-implemented cloud layer establishes an encrypted outbound connection from the on premises system, meaning no inbound firewall rules are required, and no network security posture changes are needed.
Disaster recovery and extended retention use cloud as a backup layer. Data centers take redundancy seriously in their IT infrastructure, and their physical security systems should be no different. Cloud storage can serve as a secondary retention tier for footage that needs to be preserved beyond local retention windows particularly useful for compliance scenarios where regulators or auditors may request footage going back 60, 90, or 180 days.
In CompleteView, this architecture is supported natively. The VMS runs on premises, and Salient Cloud Services connects to the deployment without disrupting how the system operates. User credentials and video data are not stored in the cloud, the cloud layer handles remote access and system management while the on premises system retains authority over the data itself. For data center operators who’ve raised concerns about cloud security, that distinction tends to resolve most of the objections.
In Part 2, we’ll get into the specific design decisions that determine whether a hybrid deployment actually performs and what integrators should expect when walking into a data center security conversation.
