Input and output reservations - AWS Elemental MediaLive

Input and output reservations

MediaLive offers input and output reservations.

Input reservation attributes and matching

An input reservation has these attributes:

  • Codec

  • Resolution (a range)

  • Bitrate (a range)

  • Region (in which the input runs)

An input reservation applies to the cost of processing input. For a reservation to apply to an input, the attributes of the input reservation must match the fields in the channel's Input specification. The channel must also run in the Region that is specified in the reservation. For example, suppose that your input specification for a channel is AVC, HD, and Max 20 Mbps. A reservation that matches those attributes could apply to the input in that channel.

Output reservation attributes and matching

An output reservation has the following attributes:

  • Codec

  • Resolution (a range)

  • Bitrate (a range)

  • Frame rate (a range)

  • Region (in which the input runs)

An output reservation applies to the cost of the processing output. For a reservation to apply to an output, the attributes of the output reservation must match the corresponding fields in the channel configuration. The channel must also run in the Region that is specified in the reservation. You can find the fields on the AWS Elemental MediaLive console:

  • For a regular video and audio output, the fields are in the Video output section of the channel configuration. To make most of the fields appear, you must choose a codec on the page.

  • For an audio-only output, the fields are in the Audio output section of the channel configuration.

How matching works

There is a match between an existing channel and a reservation if all of the fields in the channel match the corresponding reservation attributes.

If only one of the fields does not match its corresponding reservation attribute, there is no match between the output and reservation.

There is a match if the value of a field in the channel is equal to, or falls within the range of, the corresponding attribute. For example, a frame rate of 29.97 fps in the channel configuration falls within the range of a frame rate attribute of <=30fps in the reservation.

For the frame rate attribute, there is a match as follows:

  • If the channel output frame rate is set to a specific frame rate: There is a match if the frame rate specified in the channel configuration falls within the reservation frame rate range. For example, the specified frame rate is 24fps and the reservation is <=30fps.

  • If the channel output frame rate is set to initialize from the source, there is a match only if the reservation range includes 60fps. For example, there is a match on reservations with 30-60fps.

    Note: If you purchase a reservation to target a specific output, and that output has the frame rate set to initialize from source, make sure that you purchase a reservation that specifies 30-60fps. Don't purchase a reservation that specifies <=30fps.

How an input or output reservation is applied

At the start of each monthly billing cycle, AWS replenishes each reservation with the pool of minutes for the month.

At the end of the cycle, AWS applies the minutes from a given reservation to reduce the cost for the processed items (inputs or outputs) whose attributes match this reservation. For each minute in the month, AWS determines if one or more matching items were running. It accumulates these running minutes within the hour, up to a maximum of 60 minutes in the hour.

After the reservation minutes are used up for the hour, AWS charges the regular rate-per-minute for the remainder of the items in that hour.

Running minutes can be allocated over items

The running minutes could come from more than one item. For example, you start Channel A with an input that matches a given reservation. You have purchased only one instance of this reservation. After 45 minutes, you start Channel B that also has an input that matches a given reservation. After 15 more minutes, you stop Channel A. The running minutes are accumulated as shown by the shading in the following illustration.


          A graphic representation of 80 minutes with two inputs, one input running for 60
            minutes and one input running for 35 minutes. The second input has a 15 minute overlap
            with the first input. A bar runs across the top of the image and is divided into eight
            segments representing 10 minute blocks, totalling 80 minutes. Below the top bar, a block
            labelled Input is aligned underneath the first six 10 minute blocks, representing 60
            minutes beginning at the start of the total 80 minute period. This first Input block is
            fully shaded, showing it fully matches the reservation. Below the first Input block is
            another block labelled Input. This block begins after the first 45 minutes represented
            by the 10 minute blocks. On the second Input block, only the final 20 minutes, of 35,
            are shaded. The shading that represents running minutes does not overlap between the two
            Input blocks.

Here is another example of how different items can consume the running minutes. Suppose that in one hour, you run only outputs that match a given reservation. You have purchased only one instance of this reservation. You run these four matching outputs simultaneously for 15 minutes each. During that hour, you don't run any other matching outputs. Those four outputs would all contribute to the 60 minutes.


          A graphic representation of 60 minutes with four outputs, each output running for
            15 minutes. A bar runs across the top of the image and is divided into six segments
            representing 10 minute blocks, totalling 60 minutes. Below the top bar, four block are
            labelled as outputs. Each output block occupies 15 minutes of space. Each output block
            starts at the first 10 minute block. All four output blocks are fully shaded to
            represent the entire output is considered running minutes.

Processing bursts are not supported

The 60-minute rule means that reservations can't be used for processing bursts.

For example, in one hour you run four outputs that match a given reservation. You have purchased only one instance of this reservation. You run these four matching outputs simultaneously for 60 minutes each. Only one of these outputs is eligible for the reservation because one output is enough to use up the 60 running minutes per hour.


          A graphic representation of 60 minutes with four inputs, each input running for 60
            total minutes. A bar runs across the top of the image and is divided into six segments
            representing 10 minute blocks, totalling 60 minutes. Below the top bar, four block are
            labelled as inputs. Each input block occupies 60 minutes of space, running the entire
            length of the 60 minute bar. Only the first input block is shaded, representing that
            only one of the four inputs is cosuming the 60 minutes of running minutes.

Unused minutes

If some of the minutes in the reservation are not used, those minutes are lost; the minutes are not transferred to the next month.

Running minutes can be allocated over items

There are no restrictions regarding channels:

  • For example, the reservation could be consumed based on the processing of one input from one channel and another input from a different channel.

  • There is no requirement for all the inputs or outputs in a given channel to be covered by a reservation.