This topic describes the Greenplum Database 7 platform and operating system software requirements for deploying the software to on-premise hardware, or to public cloud services such as AWS, GCP, or Azure.
Greenplum Database 7 runs on the following operating system platforms:
Do not install anti-virus software of any type on Greenplum Database hosts. VMware Greenplum is not supported for use with anti-virus software because the additional CPU and IO load interferes with Greenplum Database operations.
A kernel issue in Red Hat Enterprise Linux 8.5 and 8.6 can cause I/O freezes and synchronization problems with XFS filesystems. This issue is fixed in RHEL 8.7. See RHEL8: xfs_buf deadlock between inode deletion and block allocation.
Greenplum Database server supports TLS version 1.2 on RHEL/CentOS systems, and TLS version 1.3 on Ubuntu systems.
Greenplum Database 7 requires the following software packages on RHEL systems. The packages are installed automatically as dependencies when you install the Greenplum RPM package):
VMware Greenplum Database 7 client software requires these operating system packages:
SSL is supported only on the Greenplum Database coordinator host system. It cannot be used on the segment host systems.
For all Greenplum Database host systems, if SELinux is enabled in
Enforcingmode then the Greenplum process and users can operate successfully in the default
Unconfinedcontext. If increased confinement is required, then you must configure SELinux contexts, policies, and domains based on your security requirements, and test your configuration to ensure there is no functionality or performance impact to Greenplum Database. Similarly, you should either deactivate or configure firewall software as needed to allow communication between Greenplum hosts. See Deactivate or Configure SELinux.
Greenplum Databased 7 supports these Java versions for PL/Java and PXF:
Greenplum Database uses the system default
python3 for the Greenplum management utilities, and
python3.9 for the PL/Python module. For most of the supported OS versions, the system default
python3.9. If you are installing Greenplum Database on Rocky Linux 8, the default
python3 version included is
python3.6. You may want to unify the
python3 versions to
python3.9 by running the following commands:
sudo yum install python39-psycopg2 python39-pyyaml python39-psutil sudo update-alternatives set python3 /usr/bin/python39 sudo update-alternatives set python /usr/bin/python39
VMware releases a Clients tool package on various platforms that can be used to access Greenplum Database from a client system. The Greenplum 7 Clients tool package is supported on the following platforms:
The Greenplum 7 Clients package includes the client and loader programs plus database/role/language commands and the Greenplum Streaming Server command utilities. Refer to Greenplum Client and Loader Tools Package for installation and usage details of the Greenplum 7 Client tools.
This table lists the versions of the Greenplum Extensions that are compatible with this release of Greenplum Database 7.
|Component||Package Version||Additional Information|
|PL/Java||2.0.7||Supports Java 8 and 11.|
|Python 3.9 Data Science Module Package||1.2|
|PL/R||3.1.1||(CentOS) R 3.3.3
(Ubuntu) You install R 3.5.1+.
|R Data Science Library Package||2.0.2|
|PL/Container Image for R||2.1.2||R 3.6.3|
|PL/Container Images for Python||2.1.2||Python 2.7.12
|MADlib Machine Learning||2.0.0||Support matrix at MADlib FAQ.|
|PostGIS Spatial and Geographic Objects||3.3.2|
For information about the Oracle Compatibility Functions, see Oracle Compatibility Functions.
These Greenplum Database extensions are installed with Greenplum Database
Greenplum Platform Extension Framework (PXF) - PXF provides access to Hadoop, object store, and SQL external data stores. Refer to Accessing External Data with PXF in the Greenplum Database Administrator Guide for PXF configuration and usage information.
VMware Greenplum Database versions starting with 6.19.0 no longer bundle a version of PXF. You can install PXF in your Greenplum cluster by installing the independent distribution of PXF as described in the PXF documentation.
Greenplum 5.x clients (gpload, gpfdist) are supported with Greenplum 6.x and 7x Server and Informatica PowerCenter and PowerExchange 10.4.
The following table lists minimum recommended specifications for hardware servers intended to support Greenplum Database on Linux systems in a production environment. All host servers in your Greenplum Database system must have the same hardware and software configuration. Greenplum also provides hardware build guides for its certified hardware platforms. Work with a Greenplum Systems Engineer to review your anticipated environment to ensure an appropriate hardware configuration for Greenplum Database.
|Minimum CPU||Any x86_64 compatible CPU|
|Minimum Memory||16 GB RAM per server|
|Disk Space Requirements||
|Network Requirements||10 Gigabit Ethernet within the array
NIC bonding is recommended when multiple interfaces are present
Greenplum Database can use either IPV4 or IPV6 protocols.
Resource Groups - one of the key Greenplum Database features - can control transaction concurrency, CPU and memory resources, workload isolation, and dynamic bursting.
When using resource groups to control resource allocation on Intel based systems, consider switching off Hyper-Threading (HT) in the server BIOS (for Intel cores the default is ON). Switching off HT might cause a small throughput reduction (less than 15%), but can achieve greater isolation between resource groups, and higher query performance with lower concurrency workloads.
You must run VMware Greenplum version 6.9 or later on Dell EMC DCA systems, with software version 18.104.22.168 and later.
The only file system supported for running Greenplum Database is the XFS file system. All other file systems are explicitly not supported by VMware.
Greenplum Database is supported on network or shared storage if the shared storage is presented as a block device to the servers running Greenplum Database and the XFS file system is mounted on the block device. Network file systems are not supported. When using network or shared storage, Greenplum Database mirroring must be used in the same way as with local storage, and no modifications may be made to the mirroring scheme or the recovery scheme of the segments.
Other features of the shared storage such as de-duplication and/or replication are not directly supported by Greenplum Database, but may be used with support of the storage vendor as long as they do not interfere with the expected operation of Greenplum Database at the discretion of VMware.
Greenplum Database can be deployed to virtualized systems only if the storage is presented as block devices and the XFS file system is mounted for the storage of the segment directories.
Greenplum Database is supported on Amazon Web Services (AWS) servers using either Amazon instance store (Amazon uses the volume names
ephemeral[0-23]) or Amazon Elastic Block Store (Amazon EBS) storage. If using Amazon EBS storage the storage should be RAID of Amazon EBS volumes and mounted with the XFS file system for it to be a supported configuration.
<!— VERIFY 7X interoperablity with DDBOOST
VMware Greenplum 7 supports Data Domain Boost for backup on Red Hat Enterprise Linux. This table lists the versions of Data Domain Boost SDK and DDOS supported by VMware Greenplum 7.
|VMware Greenplum||Data Domain Boost||DDOS|
|6.x||3.3||6.1 (all versions), 6.0 (all versions)|
In addition to the DDOS versions listed in the previous table, VMware Greenplum supports all minor patch releases (fourth digit releases) later than the certified version.
Greenplum Database provides access to HDFS with the Greenplum Platform Extension Framework (PXF).
PXF can use Cloudera, Hortonworks Data Platform, MapR, and generic Apache Hadoop distributions. PXF bundles all of the JAR files on which it depends, including the following Hadoop libraries:
|PXF Version||Hadoop Version||Hive Server Version||HBase Server Version|
|6.x, 5.15.x, 5.14.0, 5.13.0, 5.12.0, 5.11.1, 5.10.1||2.x, 3.1+||1.x, 2.x, 3.1+||1.3.2|
If you plan to access JSON format data stored in a Cloudera Hadoop cluster, PXF requires a Cloudera version 5.8 or later Hadoop distribution.
The operating system parameters for cloud deployments are the same as on-premise with a few modifications. Use the Greenplum Database Installation Guide for reference. Additional changes are as follows:
Add the following line to
AWS requires loading network drivers and also altering the Amazon Machine Image (AMI) to use the faster networking capabilities. More information on this is provided in the AWS documentation.
The disk settings for cloud deployments are the same as on-premise with a few modifications. Use the Greenplum Database Installation Guide for reference. Additional changes are as follows:
nobarrieroption is not supported on RHEL 8 or Ubuntu nodes.
AWS provides a wide variety of virtual machine types and sizes to address virtually every use case. Testing in AWS has found that the optimal instance types for Greenplum are “Memory Optimized”. These provide the ideal balance of Price, Memory, Network, and Storage throughput, and Compute capabilities.
Price, Memory, and number of cores typically increase in a linear fashion, but the network speed and disk throughput limits do not. You may be tempted to use the largest instance type to get the highest network and disk speed possible per VM, but better overall performance for the same spend on compute resources can be obtained by using more VMs that are smaller in size.
AWS uses Hyperthreading when reporting the number of vCPUs, therefore 2 vCPUs equates to 1 Core. The processor types are frequently getting faster so using the latest instance type will be not only faster, but usually less expensive. For example, the R5 series provides faster cores at a lower cost compared to R4.
This variable is pretty simple. Greenplum needs at least 8GB of RAM per segment process to work optimally. More RAM per segment helps with concurrency and also helps hide disk performance deficiencies.
AWS provides 25Gbit network performance on the largest instance types, but the network is typically not the bottleneck in AWS. The “up to 10Gbit” network is sufficient in AWS.
Installing network drivers in the VM is also required in AWS, and depends on the instance type. Some instance types use an Intel driver while others use an Amazon ENA driver. Loading the driver requires modifying the machine image (AMI) to take advantage of the driver.
The AWS default disk type is General Performance (GP2) which is ideal for IOP dependent applications. GP2 uses SSD disks and relative to other disk types in AWS, is expensive. The operating system and swap volumes are ideal for GP2 disks because of the size and higher random I/O needs.
Throughput Optimized Disks (ST1) are a disk type designed for high throughput needs such as Greenplum. These disks are based on HDD rather than SSD, and are less expensive than GP2. Use this disk type for the optimal performance of loading and querying data in AWS.
Cold Storage (SC1) provides the best value for EBS storage in AWS. Using multiple 2TB or larger disks provides enough disk throughput to reach the throughput limit of many different instance types. Therefore, it is possible to reach the throughput limit of a VM by using SC1 disks.
EBS storage is durable so data is not lost when a virtual machine is stopped. EBS also provides infrastructure snapshot capabilities that can be used to create volume backups. These snapshots can be copied to different regions to provide a disaster recovery solution. The Greenplum Cloud utility
gpsnap, available in the AWS Cloud Marketplace, automates backup, restore, delete, and copy functions using EBS snapshots.
Storage can be grown in AWS with “gpgrow”. This tool is included with the Greenplum on AWS deployment and allows you to grow the storage independently of compute. This is an online operation in AWS too.
Ephemeral Storage is directly attached to VMs, but has many drawbacks: - Data loss when stopping a VM with ephemeral storage - Encryption is not supported - No Snapshots - Same speed can be achieved with EBS storage - Not recommended
|Instance Type||Memory||vCPUs||Data Disks|
|Instance Type||Memory||vCPUs||Data Disks|
Performance testing has indicated that the Coordinator node can be deployed on the smallest r5.xlarge instance type to save money without a measurable difference in performance. Testing was performed using the TPC-DS benchmark.
The Segment instances run optimally on the r5.4xlarge instance type. This provides the highest performance given the cost of the AWS resources.
The two most common instance types in GCP are “Standard” or “HighMem” instance types. The only difference is the ratio of Memory to Cores. Each offer 1 to 64 vCPUs per VM.
Like AWS, GCP uses Hyperthreading, so 2 vCPUs equates to 1 Core. The CPU clock speed is determined by the region in which you deploy.
Instance type n1-standard-8 has 8 vCPUs with 30GB of RAM while n1-highmem-8 also has 8 vCPUs with 52GB of RAM. There is also a HighCPU instance type that generally isn’t ideal for Greenplum. Like AWS and Azure, the machines with more vCPUs will have more RAM.
GCP network speeds are dependent on the instance type but the maximum network performance is possible (10Gbit) with a virtual machine as small as only 8 vCPUs.
Standard (HDD) and SSD disks are available in GCP. SSD is slightly faster in terms of throughput but comes at a premium. The size of the disk does not impact performance.
The biggest obstacle to maximizing storage performance is the throughput limit placed on every virtual machine. Unlike AWS and Azure, the storage throughput limit is relatively low, consistent across all instance types, and only a single disk is needed to reach the VM limit.
Testing has revealed that while using the same number of vCPUs, a cluster using a large instance type like n1-highmem-64 (64 vCPUs) will have lower performance than a cluster using more of the smaller instance types like n1-highmem-8 (8 vCPUs). In general, use 8x more nodes in GCP than you would in another environment like AWS while using the 8 vCPU instance types.
The HighMem instance type is slightly faster for higher concurrency. Furthermore, SSD disks are slightly faster also but come at a cost.
|Instance Type||Memory||vCPUs||Data Disks|
On the Azure platform, in addition to bandwidth, the number of network connections present on a VM at any given moment can affect the VM’s network performance. The Azure networking stack maintains the state for each direction of a TCP/UDP connection in a data structures called a flow. A typical TCP/UDP connection will have 2 flows created: one for the inbound direction and another for the outbound direction. The number of network flows on Azure is limited to an upper bound. See Virtual machine network bandwidth in the Azure documentation for more details. In practice this can present scalability challenges for workloads based on the number of concurrent queries, and on the complexity of those queries. Always test your workload on Azure to validate that you are within the Azure limits, and be advised that if your workload increases you may hit Azure flow count boundaries at which point your workload may fail. VMware recommends using the UDP interconnect, and not the TCP interconnect, when using Azure. A connection pooler and resource group settings can also be used to help keep flow counts at a lower level.
Each VM type has limits on disk throughput so picking a VM that doesn’t have a limit that is too low is essential. Most of Azure is designed for OLTP or Application workloads, which limits the choices for databases like Greenplum where throughput is more important. Disk type also plays a part in the throughput cap, so that needs to be considered too.
Most instance types in Azure have hyperthreading enabled, which means 1 vCPU equates to 2 cores. However, not all instance types have this feature, so for these others, 1 vCPU equates to 1 core.
The High Performance Compute (HPC) instance types have the fastest cores in Azure.
In general, the larger the virtual machine type, the more memory the VM will have.
The Accelerated Networking option offloads CPU cycles for networking to “FPGA-based SmartNICs”. Virtual machine types either support this or do not, but most do support it. Testing of Greenplum hasn’t shown much difference and this is probably because of Azure’s preference for TCP over UDP. Despite this, UDPIFC interconnect is the ideal protocol to use in Azure.
There is an undocumented process in Azure that periodically runs on the host machines on UDP port 65330. When a query runs using UDP port 65330 and this undocumented process runs, the query will fail after one hour with an interconnect timeout error. This is fixed by reserving port 65330 so that Greenplum doesn’t use it.
Storage in Azure is either Premium (SSD) or Regular Storage (HDD). The available sizes are the same and max out at 4TB. Instance types either do or do not support Premium but, interestingly, the instance types that do support Premium storage, have a lower throughput limit. For example:
gpcheckperfto have 1424 write and 1557 read MB/s performance.
To get the maximum throughput from a VM in Azure, you have to use multiple disks. For larger instance types, you have to use upwards of 32 disks to reach the limit of a VM. Unfortunately, the memory and CPU constraints on these machines means that you have to run fewer segments than you have disks, so you have to use software RAID to utilize all of these disks. Performance takes a hit with software RAID, too, so you have to try multiple configurations to optimize.
The size of the disk also impacts performance, but not by much.
Software RAID not only is a little bit slower, but it also requires
umount to take a snapshot. This greatly lengthens the time it takes to take a snapshot backup.
Disks use the same network as the VMs so you start running into the Azure limits in bigger clusters when using big virtual machines with 32 disks on each one. The overall throughput drops as you hit this limit and is most noticeable during concurrency testing.
The best instance type to use in Azure is “Standard_H8” which is one of the High Performance Compute instance types. This instance series is the only one utilizing InfiniBand, but this does not include IP traffic. Because this instance type is n0t available in all regions, the “Standard_D13_v2” is also available.
|Instance Type||Memory||vCPUs||Data Disks|
|Instance Type||Memory||vCPUs||Data Disks|