Greenplum Database size discussed in the Planning section. You must verify that the free space in the datastore, the free memory in the cluster, and total amount of free processors in the cluster are enough build your own base virtual machine, you must have a running CentOS 7 virtual machine in the same datastore and cluster where you will deploy the Greenplum environment. You must have access to the Internet IP and Fully Qualified Domain Name (FQDN). VMware vSphere datacenter name. VMware vSphere vSAN datastore name. VMware vSphere compute cluster name. Network Request the following from your network administrator

Greenplum Database size discussed in the Planning section. You must verify that the free space in the datastore, the free memory in the cluster, and total amount of free processors in the cluster are enough build your own base virtual machine, you must have a running CentOS 7 virtual machine in the same datastore and cluster where you will deploy the Greenplum environment. You must have access to the Internet IP and Fully Qualified Domain Name (FQDN). VMware vSphere datacenter name. VMware vSphere vSAN datastore name. VMware vSphere compute cluster name. Network Request the following from your network administrator

Greenplum Database size discussed in the Planning section. You must verify that the free space in the datastore, the free memory in the cluster, and total amount of free processors in the cluster are enough build your own base virtual machine, you must have a running CentOS 7 virtual machine in the same datastore and cluster where you will deploy the Greenplum environment. You must have access to the Internet IP and Fully Qualified Domain Name (FQDN). VMware vSphere datacenter name. VMware vSphere vSAN datastore name. VMware vSphere compute cluster name. Network Request the following from your network administrator

storage (the Datastore Volume) is not dependent on the physical host. If high availability (HA) and RAID configuration are already configured at the external storage array, the Datastore Volume should the physical hosts. In the specifications below, 30% of the Datastore Volume is reserved for Greenplum operations and 30% of the Datastore Volume is reserved for vSphere operations. 10 TiB Database the operating system for the master host 1-9999 100 vSphere Datastore Operating Reservation (%) Percentage of free space in the datastore for DRS to rebalance master hosts and segment hosts across ESXi

Greenplum Database size discussed in the Planning section. You must verify that the free space in the datastore, the free memory in the cluster, and total amount of free processors in the cluster are enough decide to build your own OVA template, you must have a running CentOS 7 virtual machine in the same datastore and cluster where you will deploy the Greenplum environment. You will need it to create the virtual IP and Fully Qualified Domain Name (FQDN). VMware vSphere datacenter name. VMware vSphere vSAN datastore name. VMware vSphere compute cluster name. Network Request the following from your network administrator

system for the coordinator virtual machine 1-9999 100 vSphere Datastore Operating Reservation (%) Percentage of free space in the datastore for DRS to rebalance virtual machines across ESXi hosts 1-100 for all the virtual machines. vSphere Datastore shows the total space used for all Greenplum virtual machines, and how much space is reserved for vSphere datastore operations. Recommended Storage Size Size is the total storage you must have available in your vSphere datastore for the values you provided. Prerequisites You must meet certain prerequisites and collect specific information from your existing

.................................................................................. - 21 - 解读 GP 分布策略 .................................................................................................. ..................................................................................... - 125 - 验证分区策略 .................................................................................................. ............................................................................... - 343 - 规划 Mirror 策略 ................................................................................................. -

容器资源分配 ○ CPU ○ 内存 ○ 磁盘 ● 容器间网络互联 ○ 本机网络 ○ 跨机网络 ● 容器化Greenplum部署策略 ○ Master部署策略 ○ Primary Segment部署策略 ○ Mirror Segment部署策略 ● 容器化Greenplum运维管理 ○ 故障检测及恢复 ○ 升级扩容 ● 容器化Greenplum存储管理 ○ 容器本地存储易失性 容器资源分配 ○ CPU ○ 内存 ○ 磁盘 ● 容器间网络互联 ○ 本机网络 ○ 跨机网络 ● 容器化Greenplum部署策略 ○ Master部署策略 ○ Primary Segment部署策略 ○ Mirror Segment部署策略 ● 容器化Greenplum运维管理 ○ 故障检测及恢复 ○ 升级扩容 ● 容器化Greenplum存储管理 ○ 容器本地存储易失性 容器资源分配 ○ CPU ○ 内存 ○ 磁盘 ● 容器间网络互联 ○ 本机网络 ○ 跨机网络 ● 容器化Greenplum部署策略 ○ Master部署策略 ○ Primary Segment部署策略 ○ Mirror Segment部署策略 ● 容器化Greenplum运维管理 ○ 故障检测及恢复 ○ 升级扩容 ● 容器化Greenplum存储管理 ○ 容器本地存储易失性

4、表格设计 、表格设计 udw 的表格创建类似于 postgresql，由于 udw 采⽤ mpp 数据，创建表格的时候可以选择不同的数据分布策略，不同的存储⽅式等等。创建表格的时候可以定义下⾯信息： 数据类型 表约束 数据分布策略 表存储模型 分区策略 外部表：udwfile、udwhdfs 下⾯分别根据上⾯的可选信息对表格设计进⾏分析。 4.1 数据类型 数据类型 开发指南 Greenplum数据仓库 4.2 表约束 表约束 udw 表格⽀持 postgresql 的表格约束，拥有 primary、unique 、check、not null、foreign 等约束，主键约束必须使⽤ hash 策略来分布表数据存储，不能在同⼀个表同时使⽤主键和唯 ⼀约束，并且指定了primary 和 unique 的列必须全部或者部分包含在分布键中。 创建表检查约束 CREATE TABLE products( Hash 分布策略，并且约束列必须和表的分布键对应的列⼀致（或者是超集） CREATE TABLE products( product_no integer UNIQUE, name text, price numeric ) DISTRIBUTED BY (product_no); 主键约束：主键约束是唯⼀约束和⾮空约束的组合。要使⽤主键约束，表必须使⽤ Hash 分布策略，并且约束列

协议进行转换，之前遇到某 客户因为驱动和服务器硬件兼容问题，压力一大，服务器自就会自 动重启，最后又将网络设备改回万兆交换机。 ·网卡绑定采用 mode=4 (802.3ad)，流量传输 hash 策略选用 xmit_hash_policy=layer3+4【((sourceport XOR dest port) XOR ((source IP XOR dest IP) AND 0xffff) modulo 合 考虑服务器配置、生产环境的运行负载压力、跑批用户和前段查询用 户并发需求等各个方面。大多数场景下，4 或 6 个为宜。 同样，作为整体架构设计的重要 组成部分，ETL 服务器、监控管 理，备份策略如何规划，如何高 效组网都得在前期考虑好。在我 们的成功案例中，同一个企业级 数据平台中 Greenplum 集群和 Hadoop 集群配合运作的案例越 来越多。在中国移动的大数据架 构规范中，云化 表加载的性能 : 首先，数据的流向： Segment Host 从 gpfdist 服务请求数据，不同的 Segment Host 之间 需要互相传输数据，加载过程中数据的重分布和 Mirror 策略都需要占 用 Segment Host 之间的网络资源。 这里需要假设一下硬件指标： 主流配置，每个节点机器的磁盘 IO 性能都可达到 1GB/S 以上 (10 块 磁盘，如果 RAID 卡很优秀，甚至可以达到