The AWS Elemental MediaLive service architecture allows us to quickly scale up or down based on demand. We have optimized this to reduce the channel start time but availability of underlying physical infrastructure sometimes dictates how quickly the channel goes to the Running state. Variability in EC2 response times can cause channel start times to occasionally spike up. We proactively monitor for these spikes in start time and fix any issues that show up.
If the delay in start-up is not acceptable you could try having one or more channels pre-staged into the Running state, then replace them as they become active with content. Of course, there is additional cost associated with this approach but if you see enough traffic it might be justifiable.
Thank you for the prompt reply!
I am aware that underlying EC2 instances are created, but I really hope the times get better in the future.
I read about reserved MediaLive resources, but now we cannot handle that cost.
For people who will read this post:
I had thought of an alternative solution based on having 1-2 MediaLive channels in idle, i.e. created but not in running state.
Unfortunately, from the tests I did, the start time is the one that requires more time.
Having reserved seems to be unique solution so far.
I would like to try a reserved MediaLive channel to reduce the starting time.
I have some questions about this:
- I read that the reserved resources are paid monthly. Can we only pay for a month and then eventually cancel?
- In the MediaLive console I saw that INPUT, CHANNEL and OUTPUT resources could be reserved. To reduce the starting time, I would say that I need an INPUT and a CHANNEL reserved resource. Is it correct? Do I also need a reserved OUTPUT?
- If I select CHANNEL I only get two options: "Reserved Advanced Audio output in EU (Frankfurt)" and "Reserved Audio Normalization output in EU (Frankfurt)". What they are? What option should I choose? I thought that the channel depends on the resolution (SD, HD or UHD) and the max bitrate, instead seems like the instance is based on the audio quality type.
- Should I change something on my end? I mean when via boto3 I create a channel, AWS will handle my request and instead of creating a new channel AWS will assign the reserved resource? Or I have to perform some changes?
Edited by: fabiov90 on Mar 26, 2019 10:45 AM
Reserved instances are a 12-month commitment. You would do reservations for the combination of inputs, outputs and channels that you intend to use. Note one easy way to figure out the right reservations is to use the option in the console to match reservations to a running channel. The idea is you identify the desired configuration, then you can leave the channel in running state 24x7. If you tend to have channels running all the time this would be a way to save cost over using on-demand pricing exclusively. Take a look at https://docs.aws.amazon.com/medialive/latest/ug/reservations.html for more information on how this works...
Also, the advanced audio and audio normalization features are in case you are using either a higher-end audio codec or need to have your channel's audio levels regulated due to certain regulatory agencies.
Edited by: AWSsteve on Mar 26, 2019 10:54 AM
Hi but in my case each time I have to change the output when user streams I am constantly saving the stream in the bucket so I have to change the name of the stored stream file and for that I have to update the channel and then start the channel which takes a lot of time and my end users have to wait a long time.
- Accepted Answerasked 3 years ago
- Accepted Answerasked 2 months ago
- AWS OFFICIALUpdated a year ago
- How do I configure Lambda functions as targets for Application Load Balancers and troubleshoot related issues?AWS OFFICIALUpdated 8 months ago
- AWS OFFICIALUpdated 4 years ago
- AWS OFFICIALUpdated 2 years ago
- EXPERTpublished 2 months ago
- EXPERTpublished 2 months ago