4 comments

[ 2.7 ms ] story [ 20.5 ms ] thread
We use Goog's ProtoBuff for the same reason on Hadoop. More flexible, space and time efficient than thrift.
Thrift's compact protocol is a fairly recent addition and is both dense and fast. What are the flexibility advantages of Protocol Buffers?
Protocol Buffers are flexible because they support 'optional' fields. So we can change the structure of Protocol Buffer without breaking the system.
Thrift has the exact same feature as I discussed in my post.