[
https://issues.apache.org/jira/browse/HDDS-13378?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Wei-Chiu Chuang updated HDDS-13378:
-----------------------------------
Description:
Add a page to list the requirements and best practices of a successful
production deployment:
h1. Requirements
1. System requirements.
- Linux. Red Hat 8 / Rocky 8 or above, Ubuntu and SUSE are recommended.
- JDK 8 or above.
- Chrony or ntpd, for time synchronization.
2. High performance storage devices.
- SAS SSD or NVMe SSD are highly recommended for metadata (RocksDB and Ratis
directory)
- Hard disks are fine for DataNode data storage, but should not be used for
metadata.
- Use direct-attached storage. Do not use NAS or SAN.
3. High speed network
- Network bandwidth affects data recovery speed so use your judgement.
- Network card bandwidth 25Gbps or above is highly recommended.
- Follow the best practices of leaf-spine network topology; keep
oversubscription ratio below 3:1 to avoid network saturation.
4. Security. Optional but highly recommended.
- Kerberos environment, including KDC.
h1. Configurations
1. Linux kernel
* Make sure the Linux kernel scaling driver is in performance mode, not
powersave mode.
* Disable Transparent Hugepage.
* Disable SELinux.
* Set vm.swappiness=1.
2. Local file system
* Disable LVM for data drives.
* Use ext4 or xfs.
* Mount drives with noatime option; mount with discard option for SSD.
3. Ozone configurations
* Install Prometheus, Grafana for monitoring.
* Increase number of write pipelines allowed in the cluster:
ozone.scm.datanode.pipeline.limit and ozone.scm.ec.pipeline.minimum
* Configure sufficient heap size for OM, SCM, Recon, Datanode, S3G and HttpFs
was:
Add a page to list the requirements and best practices of a successful
production deployment:
h1. Requirements
1. System requirements.
- Linux. Red Hat 8 / Rocky 8 or above, Ubuntu and SUSE are recommended.
- JDK 8 or above.
- Chrony or ntpd, for time synchronization.
2. High performance storage devices.
- SAS SSD or NVMe SSD are highly recommended for metadata (RocksDB and Ratis
directory)
- Hard disks are fine for DataNode data storage, but should not be used for
metadata.
- Use direct-attached storage. Do not use NAS or SAN.
3. Security. Optional but highly recommended.
- Kerberos environment, including KDC.
h1. Configurations
1. Linux kernel
* Make sure the Linux kernel scaling driver is in performance mode, not
powersave mode.
* Disable Transparent Hugepage.
* Disable SELinux.
* Set vm.swappiness=1.
2. Local file system
* Disable LVM for data drives.
* Use ext4 or xfs.
* Mount drives with noatime option; mount with discard option for SSD.
3. Ozone configurations
* Install Prometheus, Grafana for monitoring.
* Increase number of write pipelines allowed in the cluster:
ozone.scm.datanode.pipeline.limit and ozone.scm.ec.pipeline.minimum
* Configure sufficient heap size for OM, SCM, Recon, Datanode, S3G and HttpFs
> [Docs] Add a Production page under Getting Started
> --------------------------------------------------
>
> Key: HDDS-13378
> URL: https://issues.apache.org/jira/browse/HDDS-13378
> Project: Apache Ozone
> Issue Type: Task
> Components: documentation
> Reporter: Wei-Chiu Chuang
> Priority: Major
>
> Add a page to list the requirements and best practices of a successful
> production deployment:
> h1. Requirements
> 1. System requirements.
> - Linux. Red Hat 8 / Rocky 8 or above, Ubuntu and SUSE are recommended.
> - JDK 8 or above.
> - Chrony or ntpd, for time synchronization.
> 2. High performance storage devices.
> - SAS SSD or NVMe SSD are highly recommended for metadata (RocksDB and Ratis
> directory)
> - Hard disks are fine for DataNode data storage, but should not be used for
> metadata.
> - Use direct-attached storage. Do not use NAS or SAN.
> 3. High speed network
> - Network bandwidth affects data recovery speed so use your judgement.
> - Network card bandwidth 25Gbps or above is highly recommended.
> - Follow the best practices of leaf-spine network topology; keep
> oversubscription ratio below 3:1 to avoid network saturation.
> 4. Security. Optional but highly recommended.
> - Kerberos environment, including KDC.
> h1. Configurations
> 1. Linux kernel
> * Make sure the Linux kernel scaling driver is in performance mode, not
> powersave mode.
> * Disable Transparent Hugepage.
> * Disable SELinux.
> * Set vm.swappiness=1.
> 2. Local file system
> * Disable LVM for data drives.
> * Use ext4 or xfs.
> * Mount drives with noatime option; mount with discard option for SSD.
> 3. Ozone configurations
> * Install Prometheus, Grafana for monitoring.
> * Increase number of write pipelines allowed in the cluster:
> ozone.scm.datanode.pipeline.limit and ozone.scm.ec.pipeline.minimum
> * Configure sufficient heap size for OM, SCM, Recon, Datanode, S3G and HttpFs
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@ozone.apache.org
For additional commands, e-mail: issues-h...@ozone.apache.org
