Martin's Boar. Chapter 6: Partitioning

What is the main reason for wanting to partition? (easy)

• Scalability

• Reliability

• Maintainability

• Durability

What is partitioning? (medium)

• Copying data between multiple nodes, where each node contains the same data

• Dividing data into parts so that each piece of data belongs to exactly one part

• A way to split data only at the level of user's application

• A way to split data only at the level of database

• A way to store data on a hard drive

How does partitioning work with replication? (medium)

• One node contains all replicas of one partition. All nodes are different.

  (Node 1 -> [A, A, A],

  Node 2 -> [B, B, B],

  Node 3 -> [C, C, C], ...)

  A, B, C, ... -> partitions

• One node contains different partitions. All nodes are different.

  (Node 1 -> [A, B, G],

  Node 2 -> [G, F, Y],

  Node 3 -> [P, E, T], ...)

  A, B, C, ... -> partitions

• One node contains ALL partitions, one copy of each. All nodes are the same.

  (Node 1 -> [A, B, C, D, ...],

  Node 2 -> [A, B, C, D, ...],

  Node 3 -> [A, B, C, D, ...],

  ...)

  A, B, C, ... -> partitions

• One node contains one partition. All nodes are different.

  (Node 1 -> [A],

  Node 2 -> [B],

  Node 3 -> [C], ...)

  A, B, C, ... -> partitions

Why is using built in hash functions a bad idea? (medium)

• They are not enough cryptograhically strong

• They don't know how to generate a hash that will be long enough (number of characters)

• Only Ruby and Java have built in hash functions

• The same input key may have a different output hash value

What is correct for 'Term-partitioned indexes'? (hard)

• Where the secondary indexes are stored in the same partition as the primary key and value

• Where the secondary indexes are partitioned separately, using indexes values.

• Only single partition needs to be updated on write

• Read of the secondary index requires a scatter/gather across all partitions

• An entry in the secondary index may include records from all partitions of the primary key

What is correct for 'Document-partitioned indexes' (hard)

• Where the secondary indexes are stored in the same partition as the primary key and value

• Where the secondary indexes are partitioned separately, using indexes values.

• Only single partition needs to be updated on write

• Read of the secondary index requires a scatter/gather across all partitions

• An entry in the secondary index may include records from all partitions of the primary key