- 117 - Greenplum Database 管理员指南 V6.2.1 版权所有：Esena(陈淼 +86 18616691889) 编写：陈淼 - 5 - 决定表是否分区的原则 .......................................................................................... - 118 - 任何与该条件有交集(互相有重叠关系)的约束都会被移除。例如存在的约束为 BETWEEN DAY 'Monday' AND DAY 'Tuesday'。那么删除'Monday'限制时会移 除整个限制。原则是有交集即移除。例如： =# ALTER ROLE generaluser DROP DENY FOR DAY 'Monday'; 这些限制登录的信息存储在 pg_catalog. 0)将会继续分为定额部分和可共享部分。创建资源组时，MEMORY_SHARED_QUOTA 属 性用于指定，为资源组配额的专享内存中，多少百分比作为可共享内存，这部分的可共 享内存，该资源组中的并发事务都可以使用，并且采用先到先得的原则来分配，正在运 行的事务可能完全不用，也可能用一些，也可能使用到全部。 MEMORY_SHARED_QUOTA 属性的取值范围是 0 到 100，缺省值是 80。 正如之前说的，CONCURRENCY

的项目，有的是从其他数据库产品迁移过来的数据模 型，有的是新设计的数据模型。无论是哪种情况，设计时请重点关注 Greenplum 的特性，要充分发挥 Greenplum 所长。 ·分布键： 均匀为第一大原则，选取更有业务意义的字段，并非必须选择原库 的主键（PK）。 ·压缩表使用： 大表都要采用压缩存储，既节省空间也节省 IO 资源。长远来看还可 降低阵列卡和磁盘的故障率。 ·行存还是列存： 列存 gpcheckcat 检查。 2) 如遇个别表对象元数据不一致的情况，通常只会影响该对象的 使用，不会影响到整个集群。如果只是个别实例中存在问题，可 以通过 Utility 模式连接到问题实例中处理。处理原则是尽量不要 直接更改系统表的数据，而是采用数据库的手段去解决，如：drop table/alter table 等。 3) persistent table 问题，这类问题往往比较棘手，影响也比较大。

应用生态。本白皮书着眼介绍了欧拉开源操作系 统平台架构、创新性及核心特点， 同时介绍了 Greenplum 作为一款深受技术爱好者喜爱的、中立的纯开源软件，践行 “Run Everywhere”原则，用全新的HTAP核心设计满足实时处理业务需求。在此也为所有为Greenplum on openEuler 成功测试运行所做努力贡献的人员表示感谢！ 摘要 Greenplum 不受限

150MB per host for Greenplum installation Approximately 300MB per segment instance for metadata Cap disk capacity at 70% full to accommodate temporary files and prevent performance degradation Network like Greenplum where throughput is more important. Disk type also plays a part in the throughput cap, so that needs to be considered too. Compute Most instance types in Azure have hyperthreading enabled

150MB per host for Greenplum installation Approximately 300MB per segment instance for metadata Cap disk capacity at 70% full to accommodate temporary files and prevent performance degradation Network like Greenplum where throughput is more important. Disk type also plays a part in the throughput cap, so that needs to be considered too. Compute Most instance types in Azure have hyperthreading enabled

150MB per host for Greenplum installation Approximately 300MB per segment instance for metadata Cap disk capacity at 70% full to accommodate temporary files and prevent performance degradation Network like Greenplum where throughput is more important. Disk type also plays a part in the throughput cap, so that needs to be considered too. Compute Most instance types in Azure have hyperthreading enabled

150MB per host for Greenplum installation Approximately 300MB per segment instance for metadata Cap disk capacity at 70% full to accommodate temporary files and prevent performance degradation Network like Greenplum where throughput is more important. Disk type also plays a part in the throughput cap, so that needs to be considered too. Compute Most instance types in Azure have hyperthreading enabled

150MB per host for Greenplum installation Approximately 300MB per segment instance for metadata Cap disk capacity at 70% full to accommodate temporary files and prevent performance degradation Network like Greenplum where throughput is more important. Disk type also plays a part in the throughput cap, so that needs to be considered too. Compute Most instance types in Azure have hyperthreading enabled

150MB per host for Greenplum installation Approximately 300MB per segment instance for metadata Cap disk capacity at 70% full to accommodate temporary files and prevent performance degradation Network like Greenplum where throughput is more important. Disk type also plays a part in the throughput cap, so that needs to be considered too. Compute Most instance types in Azure have hyperthreading enabled

150MB per host for Greenplum installation Approximately 300MB per segment instance for metadata Cap disk capacity at 70% full to accommodate temporary files and prevent performance degradation Network like Greenplum where throughput is more important. Disk type also plays a part in the throughput cap, so that needs to be considered too. Compute Most instance types in Azure have hyperthreading enabled