Hey, Recently I am trying to upgrade the managed hadoop clusters. Saw the article about Netflix using S3 as the file system instead of Hdfs.
But this would cause the lower performance due to slower IO. So could I use Hdfs and S3 together, or even other fs together? If the file is in S3 and not in hdfs, will automatically load the file. Are there any projects targeting this problem now? Thank you. Best, Gavin
