hB II Tree

II Tree

Concepts:

Node: responsible for some part of the space.

Side Pointers: the pointers to sibling nodes.

Container & Extracted Node: C has side pointers to X. Then C is container, X is extracted node.

Node-Splitting: always create a new sibling node on the same level.


Search: if target point is included in the node's directly contained space, then turn to its child node.

               else, turn to its sibling node. continue;

Search Lock: Lock held on a node is not released until the lock on the next node(child or sibling) is acquired.

                         Exclusive or Shared.


Node-Spliting: happen when a node is overflow.

                           a node is split, with part of the contents going to a new sibling node. And a new index term posted to the parent.

Example: C is split to new node X. Then C is container of X. But not forever container. C may split to another new node Y. And C be container to Y, with Y be container to X.

Index-term Posting: an index term describes the space.       Term is posted to the parent of the container.  Posted to all the parents.


Node-Consolidation: a node whose storage utilization below a threshold shall be consolidated with its CONTAINER or EXTRACTED node. usually CONTAINER.




hB-II Tree:

Concepts:

Index Node : contain a kd-tree.

Data Node: contain data records. May contain a kd-tree.

Each node stores a description of the space it is responsible for. Usaually in the form of  (LOW_VALUE, HIGH_VALUE)  for each attribute.


Splitting can happen by multi attributes. And are not symmetric unless root.


Node-Splitting:

External Marker: if split is not at the root of the kd-tree, then it is used to indicate an extract from roiginal node. (not important)

If the splitting boundary of an index node intersects the space of a child, then the child becomes multi-parent node.




Combination of two Trees:

hB-II-Tree:

replace external marker to actual address of extracted node.(side pointers)


Decoration: Help to determine the containment order of the children of a node.

                      Decorations are related to Index Nodes only, cauz data nodes do not have children.


Consolidation: consolidate a sparse node with its container node, and modify its parent node.

评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值