95. Mounting S3 as a Filesystem

Поделиться
HTML-код
  • Опубликовано: 12 дек 2024

Комментарии • 4

  • @Loige
    @Loige Год назад +4

    and we are back! ... and I have new shelves, with Ferris (the Rust language mascot) watching over my shoulder! 🦀

  • @ChrisShenton
    @ChrisShenton Год назад +1

    My team's very interested in the S3 Mountpoint, as we do a lot of work extracting metadata from large images and videos. Exiftool is written in Perl, and appears to use lots of seek() type operations to jump through media files to find the metadata. S3 Mountpoint supports this and has shown big wins for us. For a 3.3GB video file, we currently download to EC2 then run Exiftool; that takes about 25 seconds for download then 0.6 for the Exiftool extraction; with S3 Mountpoint, it only takes 0.7 seconds since it doesn't have to download. Another benefit is that we don't have to provision lots of disk on the EC2 to accommodate large video files. TL;DR I think S3 Mountpoint has huge benefits to random-access read heavy applications, especially if it's code you don't own or want to rewrite to use S3 Objects and byte-range requests.
    I'd *really* like S3 Mountpoint to be made available for Lambda and Fargate, as is EFS now. #awsWishList

    • @AWSBites
      @AWSBites  Год назад

      That's pretty amazing, Chris. 🤞hope we get your wishes for Lambda and Fargate!