- final keys
- final values
- intermediate keys
- intermediate values
- UNION DISTINCT, RANK
- OVER, RANK
- OVER, EXCEPT
- UNION DISTINCT, RANK
Q3. Rather than adding a Secondary Sort to a slow Reduce job, it is Hadoop best practice to perform which optimization?
- Add a partitioned shuffle to the Map job.
- Add a partitioned shuffle to the Reduce job.
- Break the Reduce job into multiple, chained Reduce jobs.
- Break the Reduce job into multiple, chained Map jobs.
Q4. Hadoop Auth enforces authentication on protected resources. Once authentication has been established, it sets what type of authenticating cookie?
- encrypted HTTP
- unsigned HTTP
- compressed HTTP
- signed HTTP
- Java or Python
- SQL only
- SQL or Java
- Python or SQL
Q6. To perform local aggregation of the intermediate outputs, MapReduce users can optionally specify which object?
- Reducer
- Combiner
- Mapper
- Counter
- SUCCEEDED; syslog
- SUCCEEDED; stdout
- DONE; syslog
- DONE; stdout
- public void reduce(Text key, Iterator values, Context context){…}
- public static void reduce(Text key, IntWritable[] values, Context context){…}
- public static void reduce(Text key, Iterator values, Context context){…}
- public void reduce(Text key, IntWritable[] values, Context context){…}
Q9. To get the total number of mapped input records in a map job task, you should review the value of which counter?
- FileInputFormatCounter
- FileSystemCounter
- JobCounter
- TaskCounter (NOT SURE)
- A, P
- C, A
- C, P
- C, A, P
- combine, map, and reduce
- shuffle, sort, and reduce
- reduce, sort, and combine
- map, sort, and combine
Q12. To set up Hadoop workflow with synchronization of data between jobs that process tasks both on disk and in memory, use the _ service, which is _.
- Oozie; open source
- Oozie; commercial software
- Zookeeper; commercial software
- Zookeeper; open source
- data
- name
- memory
- worker
- hot swappable
- cold swappable
- warm swappable
- non-swappable
- on disk of all workers
- on disk of the master node
- in memory of the master node
- in memory of all workers
- on the reducer nodes of the cluster
- on the data nodes of the cluster (NOT SURE)
- on the master node of the cluster
- on every node of the cluster
- distributed cache
- local cache
- partitioned cache
- cluster cache
Q18. Skip bad records provides an option where a certain set of bad input records can be skipped when processing what type of data?
- cache inputs
- reducer inputs
- intermediate values
- map inputs