Use Parse Server with S3/Cloudfront for streaming video

[dcdspace]

I am about to start a new project that requires videos to be uploaded and streamed to clients, and I'm wondering if Parse Server can integrate with AWS's Cloudfront to stream videos to clients. I'm assuming I would simply upload the video using the S3 file adapter and then integrate it with Cloudfront somehow, but not really sure how that would all work. Using AWS is a much higher learning curve than Parse which I'm already very accustomed to, so I'm hoping this can work. Thanks.

[flovilmart]

You should create an other adapter, based on S3 that would return the cloud front URL instead of the S3 URL.
Or you could with your beforeSave hooks replace the URL in the Parse.File with the cloud front URL.

[davidruisinger]

@dcdspace How did you solve this? For some reasons parse server does NOT ALWAYS store the full file path to the DB. When a key only contains 1 File, only the filename is added:

{
  "_id": "QTS1i60M7m",
  "username": "myusername",
  "email": "myuser@email.com",
  "profile_image": "79df34843495cf509a08c296394af833_QTS1i60M7m_user.jpg"
  ...
}

But as soon as I have multiple files in an array, the full file path is stored:

{
  "_id": "JDrPTXG9DJ",
  "images": [
    {
      "__type": "File",
      "name": "26934937a2eba37a28e8daaca0127c8a_image1.jpg",
      "url": "https://mybucket.s3.amazonaws.com/26934937a2eba37a28e8daaca0127c8a_image1.jpg"
    },
    {
      "__type": "File",
      "name": "818560028e8e378e6a82be06bf77ae45_image2.jpg",
      "url": "https://mybucket.s3.amazonaws.com/818560028e8e378e6a82be06bf77ae45_image2.jpg"
    }
  ]
  ...
}

So there's not really a baseURL that I can change in a hook.

[dcdspace]

I actually haven't started the Cloudfront streaming feature of my app yet, but if you've had any success I'd be happy to hear how. I like the idea mentioned above of using a beforeSave hook to simply swap out the S3 url with the Cloudfront, or make a custom file object and have a pointer to it in my object.

I'm following this tutorial: http://docs.aws.amazon.com/AmazonCloudFront/latest/DeveloperGuide/MigrateS3ToCloudFront.html

[davidruisinger]

I created a pull request for the s3 adapter: parse-community/parse-server-s3-adapter#6

This adds another option to the S3 config to specify a custom base URL for your S3 ressource (e.g. the CloudFront Url):

filesAdapter: new S3Adapter(
  "AKIAJ7YVVK5LMUCZTQFA",
  "HeyacRj1GSn3gNOLIrjlM1SOVIBBm6d291REo0fy",
  "dongxii-tokyo",
  {
    bucketPrefix: s3BucketPrefix,
    directAccess: true,
    baseUrl: 'https://images.example.com' // This could be your CloudFront URL
  }
),

[dcdspace]

Awesome thanks!

[indiandragon]

@flovilmart Parseserver+heroku+mLab was the recommended setup during initial migration days and now S3 is being recommended as a solution to all file related issues. Is the former setup not recommended anymore ? many including me have paid setup and switching to s3 immediately wouldn't be plausible.

[julianvogels]

@indiandragon MongoDB doesn't store any files, just URLs to where the file is hosted. I think AWS S3 bucket was always one the recommended platforms for file storage.