Choosing a Satellite Ingest Tier: Locked, Unlocked or Dedicated

This whitepaper turns the satellite ingest tier comparison into a procurement document. It is written for the people who approve the spend rather than operate the platform: heads of technology, operations leads, and finance stakeholders who need to understand what each tier commits them to, where the cost curves cross, and which requirements genuinely remove a choice from the table.

Author: Gleb Sazanov Role: CTO, SATLINE.TV Updated: 2026-08-04

Abstract

Satellite ingest is frequently procured as a hardware decision when it is really a commitment decision. Teams compare a monthly figure for a virtual tuner against a monthly figure for a dedicated server, conclude that the dedicated option is the professional one, and discover later that they bought rack space and an operating system rather than capability. Others take the opposite path, standardize on the cheapest virtual tier, and find themselves unable to satisfy a contract that requires a physical conditional-access module.

Three tiers are available, and they are not points on a quality scale. A locked SAT>IP tuner is a shared frequency, pinned to one satellite position and one transponder, held permanently in a receiving state. An unlocked virtual tuner is a private tuner pinned to a position but free to move within it. A dedicated server provides bare metal with PCIe DVB cards, chosen for demodulator density and for co-locating tuners with processing hardware.

This paper sets out what distinguishes them commercially, identifies the two constraints that decide the question outright, provides a weighted scorecard for the cases where judgement is required, and offers phased guidance for estates that will grow. It does not conclude that one tier is correct for everyone, because that conclusion would be false.

Executive summary

• Locked tuners suit a stable, continuously running channel lineup. They are the cheapest per frequency, they carry no acquisition delay because the lock is already held, and they cannot be retuned by accident.
• Unlocked tuners suit work where retuning is the workload: transponder discovery, network information table scanning, and programme guide collection across a position.
• Dedicated hardware suits density and co-location — many demodulators in one chassis, with conditional-access modules, encoders and storage alongside the tuners as a single managed platform.
• One constraint removes the choice: if the satellite beam does not reach your site, no quantity of local hardware will receive it. Only an IP-delivered tier can.
• One belief costs money: the assumption that conditional access requires a local tuner. Modern professional multi-CAM systems descramble an IP-delivered stream, so this is usually a design preference rather than a requirement.
• Most mature estates are mixed, and the most common efficiency is not a new tier at all — it is adding the missing frequencies to hardware that is already installed.

How to read this whitepaper

If you are specifying ingest for the first time, read sections 1 to 4 in order: they establish what each tier is, how the billing unit shapes your cost curve, what continuous operation actually requires, and the coverage constraint that can override everything else. If you already operate a platform and are evaluating an expansion, begin at section 5 and pay particular attention to sections 6 and 7, which are where unnecessary hardware purchases usually originate.

The decision framework near the end is intended for a cross-functional session rather than solo review. Ingest decisions fail most often when engineering optimizes for control, finance optimizes for the invoice, and neither party states the service obligation the platform is being built to meet.

Method and assumptions

The analysis assumes professional redistribution, monitoring or OTT use cases where ingest is expected to run continuously and to recover quickly from interruption. It assumes a channel lineup that is mostly stable, with periodic additions, and occasional need to investigate services on positions not currently carried.

Three principles are applied throughout:

• Match the billing unit to the requirement. Frequencies if the requirement is frequencies, positions if it is positions, demodulators if it is demodulators. Mismatches are where money is quietly lost.
• Separate capability from placement. Needing a GPU is not the same as needing a GPU beside the tuner. Conflating the two is the most expensive error in this category.
• Cost the operational burden, not only the invoice. Rack space, power, an operating system and a software stack are real costs that appear on no line item.

Definitions used in this comparison

Locked SAT>IP tuner means a shared frequency sold per selected frequency. Both the satellite position and the transponder are fixed when the service is ordered, and the demodulator is brought up on that carrier and held there. It is not retunable by the client, which is what allows the lock to be maintained continuously and priced below a private tuner.

Unlocked virtual tuner means a private tuner sold per satellite position. It is idle until tuned, after which any frequency visible at that position may be requested at runtime, as often as required.

Dedicated DVB server means bare-metal hardware with PCIe tuner cards installed in a Tier III facility, with dish alignment and cabling handled. Digital Devices Max SX8 Pro cards provide eight demodulators each, covering DVB-S, DVB-S2 and DVB-S2X to 256-APSK with multistream support, in configurations of one to three cards. The operating system and everything above it belongs to the customer.

1) Mission fit: what each tier is actually for

The first question is not which tier is strongest but which problem is being solved. A locked tuner solves continuous delivery of a known lineup. Because the frequency is fixed at the service level, the tuner cannot be moved off a production transponder by a misconfigured client or an automation script that walks a channel list and loses its place. In production ingest, the inability to change something is a feature.

An unlocked tuner solves discovery. Programme guide data is distributed across transponders; verifying that a service still exists where documentation claims requires visiting it; building a channel grid from nothing means tuning around a position systematically. None of this is possible on a fixed frequency, and the retune cost that is pure waste in a production lineup is simply the mechanism here.

A dedicated server solves density and coherence. Twenty-four demodulators in one chassis, with conditional-access modules, encoders and storage in the same platform, is genuinely simpler to reason about and hand over than the same capability assembled across several services. Where a headend is being built from nothing at meaningful scale, that coherence has real value.

2) The billing unit determines the cost curve

Locked tuners are sold per frequency, unlocked per satellite position, dedicated per chassis. These three units scale very differently, and the published rates make the near-term comparison straightforward.

The locked tier is consistently 26 to 29 percent below unlocked at equivalent capacity, which makes it the economical choice for a handful of known transponders. Paying the position premium buys tuning freedom, and if the plan is never to exercise that freedom, it is a premium for nothing.

The curve inverts as frequency count rises. Once a single orbital position carries many transponders you need, the position costs less than the frequencies bought individually, and unlocked becomes both cheaper and more capable. Further along, when the binding constraint is concurrent demodulators rather than distinct frequencies, a multi-card chassis changes the arithmetic again. The practical instruction is to count distinct frequencies against distinct positions before comparing headline prices, because that ratio decides the answer more reliably than the rate card does.

3) Continuity: what an already-held lock is worth

A locked tuner has completed carrier acquisition and forward-error-correction lock at provisioning and maintains that lock thereafter. A client session opening against it does not initiate a tune, because the tune is already in effect and never stopped.

This does not appear on a specification sheet as a number, and its value is invisible during normal operation. It becomes visible at the moments that matter in a continuously running service: reconnection after a network interruption, restart of streaming software, failover to a standby node. Each of those is a point at which a cold tuner would need to acquire the carrier again before a single packet moved. Degraded-signal conditions are the sharper case — a link whose margin has been reduced by weather is precisely where a cold-tuning architecture behaves worst, because acquisition can be attempted, lost, and attempted again.

For an organization whose commercial exposure includes overnight interruption, this is the single most consequential difference between the virtual tiers, and it is the reason locked is the appropriate default for production rather than merely the cheapest option.

4) Coverage: the constraint no hardware can resolve

If a satellite's footprint does not illuminate your location, or the orbital position sits below your horizon, then no dish, no tuner card and no chassis will deliver that signal. The limit is geographic rather than technical, and additional spend cannot move the building.

This is the clearest case in which the tier decision is made for you. An IP-delivered tier receives where the beam actually lands and hands the stream over the network, which makes positions and beams that are physically unavailable at your site available regardless. Coverage spans roughly 34.5°W to 80°E across around fifteen orbital positions.

The strategic implication is that virtual tiers and dedicated hardware are frequently complementary rather than competing. An operator may run a dedicated chassis for the positions visible from their own facility and locked tuners for the ones that are not. Framing the decision as an either-or misses the most common correct answer.

5) Stream delivery and where control sits

Both virtual tiers deliver the full transponder. Nothing is filtered on the customer's behalf; what crosses the link is set per session in the tuning request. Requesting pids=all delivers the complete multiplex with programme-specific information intact, equivalent to what a tuner card in your own chassis would produce from the same carrier. Requesting an explicit list transports only those elementary streams, which is how a wide-area link is kept from carrying services that would be discarded on arrival.

The commercially relevant point is where the decision sits. Filtering is a runtime parameter under customer control and changeable between sessions, not a provisioning choice made upstream. Existing analysis and demultiplexing workflows are unaffected: standard transport-stream tooling reports the full service list, tables and elementary stream identifiers exactly as it would against local hardware.

Where broadcasters distribute terrestrial multiplexes over satellite inside a T2-MI wrapper, decapsulation is performed in the reception chain, so the inner transport stream arrives with the physical layer pipe already resolved. Conventionally this requires a discrete decapsulator or gateway between demodulator and streamer — a device whose entire purpose is to remove a layer of packaging, with its own power supply, configuration and failure modes. Removing it from the architecture is a reduction in parts count, not merely a saving.

6) Conditional access: placement, not capability

A dedicated server holding tuner, conditional-access module and smartcard together is a clean, self-contained architecture and a reasonable way to build a headend where conditional access is central. It is not, however, the only way, and the belief that it is causes avoidable hardware spend.

Modern professional multi-CAM systems descramble an IP-delivered transport stream and are not bound to a tuner's interface slot. A common production arrangement takes the multiplex from a virtual tier, carries it to the operator's own facility, and descrambles there using CAM Pro modules and their smartcards. Vendor conditional-access descrambling likewise operates against SAT>IP channels, with support present in most mainstream streaming software.

The genuine exception is an older module that must physically occupy a tuner's interface slot. Those do require tuner and module in the same chassis. Establishing which category your conditional-access hardware belongs to should precede the tier decision, because it changes the answer and it is a question with a definite, checkable response.

7) Processing: needing hardware is not needing co-location

An encoder needs a card in a slot. A substantial recording archive needs real disks on a controller. Neither requirement cares how the transport stream arrived. Hardware encoding behaves identically whether the stream came across a bus or across a network.

What these workloads require is hardware under your control, which may be a server you already operate, fed from a virtual tier. Professional encode and decode acceleration is available on the larger dedicated configurations, but the same cards in your own chassis serve equally well against an IP feed.

This distinction has direct budget consequences. An operator who already owns servers, conditional-access modules and encoders, and is short a few transponders, frequently prices another chassis. The requirement in that situation is frequency coverage, not more hardware. Extra frequencies are available as an add-on to an existing dedicated server or VPS at €10 per frequency, fed into equipment already installed, against €299 per month plus setup for the smallest dedicated configuration. The difference is not marginal.

8) Reception redundancy and service continuity

Both virtual tiers are served from reception at a separate location, reached over independent network paths, so that disruption to one reception point or one route does not interrupt delivery. The facilities are Tier III with N+1 power architecture, redundant cooling and continuous monitoring.

Reproducing this independently is a project rather than a purchase: a second physical site far enough away to be genuinely uncorrelated with the first, duplicated antennas and low-noise blocks, network paths that do not share fate, and the monitoring and switching logic to present the pair as one service. Organizations that have priced the exercise understand why it rarely survives a budget review as a line item.

A single dedicated chassis has no equivalent property. It can be made redundant with a second server, preferably elsewhere, and failover logic you design, operate and periodically test — but that redundancy is yours to build and to prove. Where continuity is a contractual obligation, this should be priced explicitly rather than assumed to come with the hardware.

9) Capacity planning and how the estate scales

Capacity is limited by a different quantity in each tier, and confusing them produces over-provisioning. The working method is straightforward: enumerate the services required, collapse them to the set of distinct transponders that carry them, then collapse that to distinct orbital positions. Services sharing a transponder need no additional tuner.

The ratio of frequencies to positions is the most useful figure in the whole exercise. Where frequencies substantially outnumber positions, the position purchase becomes favorable even if tuning freedom is never used. Where concurrent demodulator count is binding instead, the chassis arithmetic applies.

Bandwidth should be sized against the request you intend to make, not against a nominal figure. A full multiplex request must be budgeted at the transponder's full rate; an explicit stream selection is budgeted at the selected services plus table overhead. Sizing against the wrong one of these is a common and avoidable planning error.

10) Operational burden and the costs that appear nowhere

The dedicated tier's monthly figure understates its true cost. Rack space and power are consumed continuously. The operating system, streaming platform, configuration, updates and monitoring are the customer's to maintain, and root administration implies somebody on call.

One frequently cited objection does not hold: the tuner cards present as standard operating-system DVB devices and are detected automatically by compatible software, with no vendor-specific drivers and no integration work at the driver layer. The weight is not the hardware. It is ownership of the entire software stack above it.

By contrast, the virtual tiers place the reception chain, decapsulation and the platform beneath the SAT>IP interface with the provider. Client compatibility is broad rather than proprietary, since the interface is standards-compliant. For organizations whose engineering capacity is better spent on their own product than on maintaining an ingest platform, that division of labour is the substantive argument, and it is an organizational argument rather than a technical one.

Decision framework

Begin with the two questions that can remove the choice entirely, because neither is a matter of judgement:

• Is every position you need visible from your own site? If not, an IP-delivered tier is required for those positions regardless of any other consideration.
• Does any conditional-access module have to occupy a tuner's interface slot? If yes, those services require a chassis. If the modules descramble an IP-delivered stream, they do not.

Where both questions leave the field open, score five dimensions from 1 to 5 and weight them with commercial input rather than engineering preference alone:

• Lineup stability — how often the frequency list has actually changed in the past year.
• Continuity exposure — what an interruption at three in the morning costs the business.
• Frequency-to-position ratio — the count derived in section 9.
• Existing hardware — servers, modules and encoders already installed and paid for.
• Operational capacity — whether a team exists to own an operating system and software stack.

High stability and high continuity exposure point to locked. A low frequency-to-position ratio with frequent change points to unlocked. High demodulator density with genuine co-location benefit points to dedicated. Substantial existing hardware with a small shortfall points to add-on frequencies, and that combination is more common in practice than any of the three headline answers.

Phased guidance

Phase 1 — establish and measure (0–6 months). Place the production lineup on locked frequencies and retain one unlocked tuner for scanning, guide collection and diagnostics. Instrument reconnection behavior and record how often the lineup actually changes, since that measurement drives every later decision and is usually assumed rather than known.

Phase 2 — consolidate on evidence (6–18 months). Revisit the frequency-to-position ratio with real data. Positions that have accumulated many required transponders are candidates for conversion to unlocked. Positions with one or two stable transponders should remain locked. Where a genuine co-location requirement has emerged, scope a chassis to that requirement specifically rather than migrating the estate.

Phase 3 — scale deliberately (18 months onward). Add capacity in the unit that is actually constrained. Where hardware is already installed and the shortfall is coverage, add frequencies rather than chassis. Keep diagnostic capability separate from production so that a scanning operation can never become an outage.

Practical rule: buy the unit you are short of. Most unnecessary ingest spend comes from purchasing a chassis when the actual shortfall was a frequency, or purchasing a position when the actual need was one transponder.

Conclusion

Locked, unlocked and dedicated tiers are not stages of seriousness. They answer different questions, and the correct architecture for most operators uses more than one of them. Locked tuners carry production because a fixed, pre-warmed, redundantly received frequency is what production wants. An unlocked tuner earns its premium by making discovery possible. Dedicated hardware earns its place where density and co-location genuinely pay, and nowhere else.

Two errors account for most wasted spend in this category. The first is treating a dedicated chassis as the professional default, when its real advantages are narrow and its operational burden is broad. The second is assuming that a capability requirement — conditional access, encoding, recording — implies a placement requirement. It usually does not.

For an answer-first summary suitable for circulation, use the paired documentation version. For procurement and architecture review, use this whitepaper as the baseline and revisit the scorecard whenever the lineup, contract terms or installed hardware change materially.