345 Configuring Resource Groups 346 Low Memory Queries 346 Administrative Utilities and admin_group Concurrency 347 Memory and Resource Management with Resource Queues 347 Memory and Resource Management Instances 386 Best Practices 386 Additional Information 386 Segment Mirroring Configurations 387 Group Mirroring 388 Spread Mirroring 388 Block Mirroring 389 Implementing Block Mirroring 390 Overview Groups 1017 Using Resource Groups 0 Understanding Role and Component Resource Groups 1017 Resource Group Attributes and Limits 1018 Memory Auditor 1019 Transaction Concurrency Limit 1020 CPU Limits 1020

336 Configuring Resource Groups 337 Low Memory Queries 337 Administrative Utilities and admin_group Concurrency 338 VMware Greenplum v6.18 Documentation VMware, Inc. 15 Memory and Resource Management Instances 377 Best Practices 377 Additional Information 377 Segment Mirroring Configurations 378 Group Mirroring 379 Spread Mirroring 379 Block Mirroring 380 Implementing Block Mirroring 381 Overview Groups 1008 Using Resource Groups 0 Understanding Role and Component Resource Groups 1008 Resource Group Attributes and Limits 1009 Memory Auditor 1010 Transaction Concurrency Limit 1011 CPU Limits 1011

1185 Using Resource Groups 1186 Understanding Role and Component Resource Groups 1186 Resource Group Attributes and Limits 1187 Memory Auditor 1188 Transaction Concurrency Limit 1189 CPU Limits 1189 1195 memory_spill_ratio and Low Memory Queries 1195 About Using Reserved Resource Group Memory vs. Using Resource Group Global Shared Memory 1195 Other Memory Considerations 1196 Using VMware Greenplum Resource Group to a Role 1202 Monitoring Resource Group Status 1203 Viewing Resource Group Limits 1203 Viewing Resource Group Query Status and CPU/Memory Usage 1203 Viewing Resource Group CPU/Memory

272 Configuring Resource Groups 272 Low Memory Queries 273 Administrative Utilities and admin_group Concurrency 273 Memory and Resource Management with Resource Queues 273 Memory and Resource Management Instances 312 Best Practices 312 Additional Information 312 Segment Mirroring Configurations 313 Group Mirroring 314 Spread Mirroring 314 Block Mirroring 315 Implementing Block Mirroring 316 Overview Groups 942 Using Resource Groups 0 Understanding Role and Component Resource Groups 943 Resource Group Attributes and Limits 944 Memory Auditor 944 Transaction Concurrency Limit 945 CPU Limits 945

interval, float MEDIAN (expression) Example: SELECT departmzent_id, MEDIAN(salary) FROM employees GROUP BY department_id; Can take a two-dimensional array as input. Treats such arrays as matrices. sum(array[]) (label[], label, expr) int[], bigint[], float[] pivot_sum( array[‘A1’,‘A2’], attr, value) A pivot aggregation using sum to resolve duplicate entries. unnest (array[]) set of anyelement unnest( array[‘one’ 1133 Using Resource Groups 1134 Understanding Role and Component Resource Groups 1135 Resource Group Attributes and Limits 1136 Memory Auditor 1136 Transaction Concurrency Limit 1137 CPU Limits 1137

interval, float MEDIAN (expression) Example: SELECT departmzent_id, MEDIAN(salary) FROM employees GROUP BY department_id; Can take a two-dimensional array as input. Treats such arrays as matrices. sum(array[]) (label[], label, expr) int[], bigint[], float[] pivot_sum( array[‘A1’,‘A2’], attr, value) A pivot aggregation using sum to resolve duplicate entries. unnest (array[]) set of anyelement unnest( array[‘one’ Resources 989 Using Resource Groups 990 Understanding Role and Component Resource Groups 991 Resource Group Attributes and Limits 991 Memory Auditor 992 VMware Tanzu Greenplum 6 Documentation VMware, Inc

Views 973 Using Resource Groups 974 Understanding Role and Component Resource Groups 974 Resource Group Attributes and Limits 974 VMware Greenplum 7 Documentation VMware by Broadcom 43 Transaction Concurrency Assigning a Resource Group to a Role 986 Monitoring Resource Group Status 986 Viewing Resource Group Limits 987 Viewing Resource Group Query Status 987 Viewing Resource Group Memory Usage Per Host Host 987 Viewing the Resource Group Assigned to a Role 987 Viewing Resource Group Disk I/O Usage Per Host 987 Viewing a Resource Group’s Running and Pending Queries 987 Cancelling a Running or Queued Transaction

341 Configuring Resource Groups 341 Low Memory Queries 341 Administrative Utilities and admin_group Concurrency 342 Memory and Resource Management with Resource Queues 342 Resolving Out of Memory Instances 381 Best Practices 381 Additional Information 381 Segment Mirroring Configurations 382 Group Mirroring 382 Spread Mirroring 383 Block Mirroring 384 Implementing Block Mirroring 385 Overview 1006 Using Resource Groups 1007 Understanding Role and Component Resource Groups 1008 Resource Group Attributes and Limits 1009 Memory Auditor 1010 Transaction Concurrency Limit 1010 CPU Limits 1011

在 Hash join 通 过 COST 选择小表做 Hash，在 NestloopJoin 中选择 index 提高 join 性 能等等。 MPP 数据库对于 Aggregation（聚合）提供 Multiple-agg、Group- agg、sort-agg 等多种技术来提供计算性能；Mapreuce 需要开发人 员自己实现。 另外，Mapreduce 在整个 MAP->Shuffle->Reduce 组（MC-LAG Multi-ChassisLink Aggregation Group） 的 方 式将两台交换机连在一起，在服务器上将网卡通过 LACP（IEEE 802.3ad/802.1ax Link Aggregation Control Protocol）协议做绑定形 成链路聚合组 (LAG Link Aggregation Group)，如下图所示： Big Date2.indd nspname='public') and n.nspname not like 'pg_tmp_%' and c.oid not in(selectparrelid from pg_partition) group by 1 havingmax(c.tuples)/avg(c.tuples)>2 ormin(c.tuples)/avg(c.tuples)<0.5 这种方法的优势是，速度快，至于其中的几个 Magic

Role and Component Resource Groups 1171 VMware Greenplum 6 Documentation VMware, Inc 51 Resource Group Attributes and Limits 1172 Memory Auditor 1173 Transaction Concurrency Limit 1174 CPU Limits 1174 1179 memory_spill_ratio and Low Memory Queries 1180 About Using Reserved Resource Group Memory vs. Using Resource Group Global Shared Memory 1180 Other Memory Considerations 1181 Using VMware Greenplum Resource Group to a Role 1186 Monitoring Resource Group Status 1187 Viewing Resource Group Limits 1187 Viewing Resource Group Query Status and CPU/Memory Usage 1187 Viewing Resource Group CPU/Memory