公开(公告)号：US09501526B2

公开(公告)日：2016-11-22

申请号：US13865082

申请日：2013-04-17

Applicant: Yahoo! Inc.

Inventor： Rick James

IPC: G06F17/30

CPC classification number: G06F17/30486 ,  G06F17/30241 ,  G06F17/30442

Abstract: Systems and methods are disclosed for efficiently searching a database including a table. The table is partitioned along a first dimension, such as latitude. Indexes in each of the partitions are built along a second dimension, such as longitude. The table may also be clustered according to an index to improve performance. When a processor receives a database query term and a starting location, the processor selects, from the table, a data partition within a first offset to the starting location along the first dimension. The processor selects, from the selected data partition, a first number of first data records related to the query term and within a second offset to the starting location along the second dimension.

Abstract translation: 公开了用于有效地搜索包括表的数据库的系统和方法。 表沿第一维度（如纬度）进行分区。 每个分区中的索引都沿着第二个维度（如经度）构建。 还可以根据索引对表进行聚类以提高性能。 当处理器接收到数据库查询项和起始位置时，处理器从表中选择沿着第一维度到起始位置的第一偏移量内的数据分区。 所述处理器从所选择的数据分区中选择与所述查询项相关的第一数量的第一数据记录，并且沿着所述第二维度在到所述起始位置的第二偏移量内。

公开(公告)号：US20150032720A1

公开(公告)日：2015-01-29

申请号：US13948610

申请日：2013-07-23

Applicant: Yahoo! Inc.

Inventor： Rick James

IPC: G06F17/30

CPC classification number: G06F16/2453 ,  G06F16/21

Abstract: A system and method for database management and accessing data stored in the database may be optimized based on the database structure. For example, data associated with a range (e.g. owners of IP addresses) may have query requests optimized by utilizing two tables. The first table may include all the ranges, such that the ranges are continuous and not overlapping, and an identifier for each of the ranges. The second table may include the identifiers and the data that is associated with the range corresponding with those identifiers. Query requests for a particular range may be optimized by utilizing specialized database queries for the first table and because the data associated with the range is not part of the first table. Further, the second table size is reduced because an identifier may be associated with multiple ranges.

Abstract translation: 可以基于数据库结构优化用于数据库管理和访问数据库中存储的数据的系统和方法。 例如，与范围相关联的数据（例如IP地址的所有者）可以通过利用两个表来优化查询请求。 第一表可以包括所有范围，使得范围是连续的并且不重叠，以及每个范围的标识符。 第二表可以包括与与这些标识符相对应的范围相关联的标识符和数据。 可以通过对第一个表使用专门的数据库查询来优化特定范围的查询请求，并且因为与该范围相关联的数据不是第一个表的一部分。 此外，由于标识符可以与多个范围相关联，所以第二表大小减小。

公开(公告)号：US20150120745A1

公开(公告)日：2015-04-30

申请号：US14064707

申请日：2013-10-28

Applicant: Yahoo! Inc.

Inventor： Rick James

IPC: G06F17/30

CPC classification number: G06F17/30306 ,  G06F17/30315

Abstract: A system and method for database storage may be dynamically optimized for entity-attribute-value (EAV) data. In one embodiment, this dynamic EAV database may include one column for the entity identification (“ID”) and a second column for the serialized data for that ID (each attribute and the values for those attributes). The database may be dynamically updated to move or copy certain attributes to a separate column. The dynamic updating of one or more attributes may be based on certain factors (seasonal, temporal, usage, etc.), such as frequent requests for a particular attribute. This dynamic updating and indexing of particular attributes may improve certain queries and data requests.

Abstract translation: 用于数据库存储的系统和方法可以针对实体属性值（EAV）数据进行动态优化。 在一个实施例中，该动态EAV数据库可以包括用于实体标识（“ID”）的一列和用于该ID的序列化数据的第二列（每个属性和这些属性的值）。 可以动态地更新数据库以将某些属性移动或复制到单独的列。 一个或多个属性的动态更新可以基于某些因素（季节，时间，使用等），例如对特定属性的频繁请求。 特定属性的动态更新和索引可以改进某些查询和数据请求。

公开(公告)号：US20140317086A1

公开(公告)日：2014-10-23

申请号：US13865082

申请日：2013-04-17

Applicant: YAHOO! INC.

Inventor： Rick James

IPC: G06F17/30

CPC classification number: G06F17/30486 ,  G06F17/30241 ,  G06F17/30442

Abstract: Systems and methods are disclosed for efficiently searching a database including a table. The table is partitioned along a first dimension, such as latitude. Indexes in each of the partitions are built along a second dimension, such as longitude. The table may also be clustered according to an index to improve performance. When a processor receives a database query term and a starting location, the processor selects, from the table, a data partition within a first offset to the starting location along the first dimension. The processor selects, from the selected data partition, a first number of first data records related to the query term and within a second offset to the starting location along the second dimension.

Abstract translation: 公开了用于有效地搜索包括表的数据库的系统和方法。 表沿第一维度（如纬度）进行分区。 每个分区中的索引都沿着第二个维度（如经度）构建。 还可以根据索引对表进行聚类以提高性能。 当处理器接收到数据库查询项和起始位置时，处理器从表中选择沿着第一维度到起始位置的第一偏移量内的数据分区。 所述处理器从所选择的数据分区中选择与所述查询项相关的第一数量的第一数据记录，并且沿着所述第二维度在到所述起始位置的第二偏移量内。