You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
* Introducing new SHUFFLE_MAP type for bypassMerge sort shuffle
* Adding numOfPartitions as part of SGX RDD (needed by the shuffle partitioner)
* Introducing SGX partitioner - hash-based for now (to be changed by any pseudorandom function we want)
* Introducing sgxWrite interface as part of Shuffle writter
When SGX is enable we use this path instead
The method takes as arguments both the encrypted records AND the record-partition mapping (recordMapping) as return by the enclave worker
* Handling SHUFFLE_MAP_BYPASS task type in SGX worker - return a record-partition mapping iterator
* ShuffleMapTask checks if SGX is enabled and calls the appropriate shuffle interface
* ByPassSort shuffle test case going through SGXWorker now working #15
Multiple partitions supported
This is the default hash-ish shuffle (supported by SparkSQL) using serialized objects when there is no need for map-side aggregation.
Part of #9
The text was updated successfully, but these errors were encountered: