We've recently released a set of performance and stability improvements to our VOD Asset ingestion pipeline. These improvements will apply to all new Assets.
Although we have retry mechanisms in place to ensure all assets are eventually processed successfully even when we hit temporary networking or other issues, having to begin processing an asset again comes with a cost in terms of the total "time to ready".
By rewriting the part of the pipeline that processes input files before sending them to be transcoded, we've reduced this retry rate from 7% of assets to less than 0.1%.
As part of the same rewrite, we've also increased the amount of parallelisation we're able to take advantage of and decreased the duration of this pre-processing step from several minutes for an hour-long file to under 30 seconds.