In a cluster of Fusioninsight HD V100R002C60, which of the following components need to plan metadata partition? (Multiple choice)

Question

Options

• AHDFS
• BZookeeper
• CStreaming
• DRedis
• EHBase
• FKafka

• A
  79% (23)
• D
  10% (3)
• E
  3% (1)
• F
  7% (2)

Explanation

HDFS, ZooKeeper, and Streaming are the three components that require metadata partition planning because they each maintain critical cluster-level metadata that must be deliberately distributed across partitions to ensure high availability and fault tolerance. HDFS's NameNode holds the entire filesystem namespace and block mapping metadata, making its partition plan essential. ZooKeeper stores coordination metadata (leader election results, service registrations, configuration), which must be partitioned across an odd number of nodes (quorum). FusionInsight Streaming maintains topology and task-scheduling metadata that also requires explicit partition planning.

Why the distractors are wrong:

• Redis (D): Redis is a key-value cache/data store - it holds application data, not cluster metadata that requires partition planning at the infrastructure level.
• HBase (E): HBase offloads its metadata (region locations, master info) into ZooKeeper, so its metadata partition needs are already covered by ZooKeeper's plan, not separately planned.
• Kafka (F): Kafka similarly stores its topic/partition metadata in ZooKeeper rather than managing an independent metadata partition of its own.

Memory tip: Think of it as the "infrastructure backbone trio" - HDFS (storage layer), ZooKeeper (coordination layer), and Streaming (processing topology layer) each own their own metadata stores. HBase and Kafka are consumers of ZooKeeper's metadata service, and Redis is a pure data cache - none of the three need their own metadata partition plan.

No community discussion yet for this question.