1 個回答
- 最新
- 最多得票
- 最多評論
0
In PySpark you can use forEachPartition and call Kinesis or any external API for that partition, or you can also use map and call out to Kinesis per record.
# Approach 1: Per Partition
def pushToKinesis(iterator):
print(list(iterator)[0]
#push to kinesis using boto3 APIs
rdd.foreachPartition(pushToKinesis())
# Approach 2: Per record
def pushToKinesis(record):
#push to kinesis using boto3 APIs
rdd.map(lambda l: pushToKinesis(l))
The blog post below uses Approach 2 to call Amazon Comprehend:
已回答 6 年前
相關內容
- AWS 官方已更新 1 年前