AWS弹性transcorder或运行在EC2上的ffmpeg？更便宜？ [英] AWS elastic transcorder or running ffmpeg on EC2 ? Which is cheaper?

问题描述

我有一个分配转换大集视频转换为OGG，WEBM和MP4，针对不同的分辨率加上手机等，我正在考虑使用的ffmpeg或使用弹性跨codeR运行转码作业队列。 （背景信息：的源文件已经在S3和Web应用程序是相当整合到AWS SDK和文件每日得到补充，不要求实时转换）

I have an assignment to convert a large set of videos into ogg, webm and mp4, for different resolutions plus mobile etc, I am considering running transcoding job queues using ffmpeg or use elastic transcoder. (Background info: the source files are already on S3 and the web application is quite integrated into AWS SDK, and files get added daily, real time conversion not required)

我知道使用EC2和ffmpeg的会涉及到一些开发时间和精力，因为只用transcorder API反对。但是，如果有节约成本的好...

I know using EC2 and ffmpeg will involve some dev time and effort, as opposed to just using the transcorder API. But if there is good cost saving...

所以我有两个问题。

1）AWS说收费是视频（SD）大约$ .015 /分。我的问题是，是否另行收取每个输出文件。文档说，他们支持多达30输出同样的工作。但尚不清楚是否是每个工作的分钟或每每个输出格式（视频$ 0.45 /分钟30的输出格式的话）？

1) AWS say charge is roughly $.015/minute of video(SD). My question is, is each output file charged separately. Docs say they support up to 30 outputs for the same job. But it not clear whether it is per minute of the job or per minute of each output format ($0.45/minute of video for 30 output formats then) ?

2）没有任何一个有使用EC2和ffmpeg的经验，在这个用例？比transcorder这个便宜？

2) Does any one have experience using EC2 and ffmpeg in this use case? Is this cheaper than transcorder??

我上次开发一些这种性质的，我并没有为EC2和我的I / O成本因素/ O结束了耗资超过本身的实例。是否有任何code优化，我可以做，使之更便宜？ （除了使用现货实例，更便宜的地区等）。

Last time I developed something of this nature, I did not factor in the I/O costs for EC2 and I/O ended up costing more than instance itself. Are there any code optimizations that I can do, to make it cheaper? (apart from using spot instances, Cheaper Regions etc).

推荐答案

AWS弹性转码服务只支持MP4 / H.264 / AAC。它目前不支持的WebM / VP8 / Vorbis格式或Ogg / Theora格式/ Vorbis格式（这是你要找的另外两个容器/视频/音频的组合）。

AWS Elastic Transcoding Service only supports MP4/H.264/AAC. It does not currently support WebM/VP8/Vorbis or Ogg/Theora/Vorbis (which are the other two container/video/audio combinations you're looking for).

因此​​，一个自定义的ffmpeg安装为目的的要求您的任务的，但你可以非常快速地运行误入歧途的法律问题，如果你使用的软件库的错误组合。它是专利法问题的法律雷区。

As such, a custom ffmpeg installation is a requirement for the purpose of your assignment, but you can very quickly run astray of legal issues if you use the wrong combination of software libraries. It's a legal minefield of patent law issues.

一旦你过去，你需要弄清楚你在找什么。廉价和放大器;慢与快＆放大器;昂贵。经前建成这事，我会建议什么比一个EC2实例XL小。如果您尝试使用微型或小，是ppared花几个小时等待周围的编码完成$ P $。

Once you get past that, you need to figure out what you're looking for. Cheap & Slow vs. Fast & Expensive. Having built this very thing before, I would recommend nothing smaller than an EC2 XL instance. If you try to use a Micro or a Small, be prepared to spend hours and hours waiting around for the encoding to complete.

你还需要考虑如何触发EN codeS上载，如果这就是你要的东西。 ETS并没有做到这一点 - 你必须手动触发它们

You'll also need to think about how to trigger the encodes on upload, if that's what you're going for. ETS doesn't do that — you have to trigger them manually.

这篇关于AWS弹性transcorder或运行在EC2上的ffmpeg？更便宜？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！