CDN Deflection (60–90%)

Rilla can achieve 60-90% CDN deflection in suitable live conditions, reducing CDN delivery load where peer density and network quality support it.

Deflection turns delivery into a measurable outcome

CDN deflection is the percentage or volume of played video bytes delivered through Rilla rather than directly from CDN infrastructure.

Deflection connects product behavior to business impact: it shows whether peak audience demand is being absorbed by Rilla instead of only increasing CDN-served traffic.

Measurement uses real playback traffic, not theoretical capacity.

CDN and peers work in parallel

Rilla uses a hybrid delivery model: CDN and peer delivery work in parallel. This model reduces CDN-served traffic only when peer delivery is suitable.

Delivery path	Role
Your CDN	Primary delivery mechanism, instant fallback path, and delivery path for traffic that is not eligible for peer assistance.
Peer-assisted delivery	Additional delivery path where viewers relay video segments to others watching the same stream.
AI Orchestrator	Scores peers and assigns delivery relationships based on real-time conditions.
Performance P2P SDK	Executes segment-level delivery decisions inside the player.

Less CDN traffic without replacing the CDN

When a viewer requests video segments, Rilla evaluates whether those segments are eligible for peer-assisted delivery. The SDK coordinates with the Orchestrator, attempts peer delivery when conditions support it, and keeps CDN fallback available.

Traffic stays on your CDN when peer delivery is unsuitable, when no eligible peers are available, or when operational rules require a request to bypass Rilla.

This is why deflection is not the same as replacing your CDN. Rilla reduces duplicate CDN traffic where it can do so safely, while your CDN remains central to the delivery architecture.

Deflection rate impacts

Deflection rate depends on how much eligible playback traffic can be matched across viewers without crossing QoE guardrails. The strongest rollout candidates combine audience overlap, consistent media requests, eligible platforms, and network conditions that support reliable peer delivery.

Category	Considerations
Audience and network conditions	Strongest when many viewers watch the same live or linear segments at roughly the same time. Regional density, ISP concentration, NAT behavior, peer throughput, and fallback frequency affect how much traffic can safely move through peers. Start where audience overlap is predictable and enough viewers can connect locally.
Delivery architecture	Limited to eligible platforms where the Performance P2P SDK is deployed and to requests that are eligible for peer delivery. DRM license requests, authentication, entitlement, playlist access, analytics, and other CDN-only or bypassed requests stay outside the deflection expectation. Start with one high-volume eligible platform and a clear request-eligibility model.
Playback and content strategy	Improves when viewers request the same renditions, segment durations, and packaging profile. Heavy rendition fragmentation, highly volatile ABR behavior, or frequent quality switching can reduce segment-level matching. Start with content profiles that have stable ABR behavior and enough viewers on common renditions.
Ad strategy	Shared content windows usually produce cleaner deflection than highly personalized ad breaks. SSAI, targeted ads, ad markers, and ad-break personalization can reduce shared segment overlap or require request-level bypass rules. Start where media overlap is easiest to reason about, then validate how personalized ad breaks affect deflection.
Rollout and measurement design	Clean cohorts, CDN-only controls, comparable event windows, and QoE deltas make the commercial result credible. Lower deflection can be the correct result when CDN fallback protects and enhances playback. Start where baseline traffic, peer-delivered GB, CDN-delivered GB, fallback behavior, startup time, and rebuffering can be measured clearly.

Capacity, cost, and quality stay connected

Deflection connects Rilla’s commercial outcomes and makes those outcomes measurable during POCs:

Outcome	How deflection contributes
Capacity	Peer-assisted paths can absorb traffic that would otherwise rely only on CDN capacity.
Cost	Fewer CDN-served bytes can improve event-level delivery economics.
Quality	Deflection is constrained by QoE guardrails and CDN fallback, not pursued at any cost.

Measuring deflection safely

Deflection measurement combines delivery and playback metrics. Rilla-enabled cohorts are compared against CDN-only control groups before deflection results inform broader rollout planning.

Metric	What it shows
CDN deflection rate	Percentage of eligible played bytes delivered through Rilla.
Peer-delivered GB	Volume shifted through the peer-assisted layer.
CDN-delivered GB	Remaining CDN traffic after deflection.
Deflection during peak windows	Performance under the highest live-event demand.
P2P connection success rate	Whether peer delivery was viable.
Fallback frequency	How often CDN protection was required.
Startup and rebuffer deltas	Whether deflection stayed within QoE guardrails.

Cut Delivery Cost How deflection maps to CDN-served traffic and event economics.

Capacity on demand How peer-assisted paths add capacity during high concurrency.

Quality Under Load Why deflection is measured with QoE guardrails.

Architecture Technical view of the hybrid CDN and peer delivery model.