Deploy Dell PowerFlex v3.6.x

Prepare the CSV file

Create a CSV topology file to deploy PowerFlex with the PowerFlex Installer on Linux or Windows-based nodes. Follow the instructions that match your environment and the PowerFlex support matrix for your version. CSV file templates are provided as part of the software download, in the PowerFlex Gateway software packages. They are also accessible from the PowerFlex Installer user interface.

1. In the CSV file, fill in your site-specific information in the appropriate places, overwriting the default information provided in the file.
   You only need to use one spreadsheet for the installation, as follows:
   • To manually enter all configuration details, use the Complete spreadsheet.
   • To use default values for the non-mandatory fields, use the Minimal spreadsheet.
   • To configure non-default values for columns that are not in the Minimal spreadsheet, either use the Complete spreadsheet, or copy the column heading from there into the Minimal spreadsheet, and enter your custom values into the Minimal spreadsheet.
   The following figure shows part of a sample Complete CSV file used to deploy Linux nodes in a hyper-converged system:

   
   
   

   The following figure shows a close up of a sample Complete CSV file, where SDCs are defined below the rows for MDMs and SDSs. In a two-layer configuration, the Linux servers will form the MDM cluster, and each of those servers will also be SDSs. SDCs can be added here, in separate rows, or added to the system later.

   
   
   

2. For Linux nodes, the Domain column is not relevant. Leave the column blank or remove the column.
   The following table describes the fields in the spreadsheets. The required fields appear in both complete and minimal spreadsheets. Field names are case-sensitive; the order of the columns is not significant.

   Table 1. CSV topology spreadsheets

   Field	Description	Required
   Domain	If using a domain user, the name of the domain (not relevant for Linux)
   Username	The name of the user. Linux nodes: Enter either root or a non-root sudo user, indicated by appending "(sudo)" to the end of the username (default: root). For more information on deploying with a non-root user, see "Deployment of a non-root user". Windows nodes: Enter a user with administrator rights (default: administrator).
   Password	Password used to log in to the node. This should be the password of the user entered in the Username column. To authenticate with SSH instead of node passwords, see "Using SSH authentication on the PowerFlex Gateway.	Yes
   Operating System	The server’s operating system: linux or windows	Yes
   IPs	IP address to be used for multiple purposes. Use this field to designate one IP address that will be assigned to all of the following: MDM IP, MDM Mgmt IP and SDS All IP. This option is provided for use cases where separate networks for data and management are not required.
   Is MDM/TB	The MDM role to deploy on this node: Primary, Secondary, TB, Standby-Secondary, Standby-TB, or blank (if not an MDM). For more information, see "The MDM cluster" in the Architecture section of Getting to Know PowerFlex.	Yes
   MDM Mgmt IP	The IP address for the management-only network. The management IP address is not required for: Tiebreaker, Standby Tiebreaker, and any host that is not an MDM. In such cases, leave this field blank.
   MDM IPs	MDM IP addresses used to communicate with other PowerFlex components in the storage network. This is required for all MDMs, Tiebreakers and Standbys. Leave this field blank for hosts that are not part of the MDM cluster. Maximum of eight addresses, comma-separated, no spaces.
   MDM Name	The name to be assigned to an MDM node, up to 31 characters in length. MDM names must be unique and may not contain spaces. Ensure that each server on which MDM is installed has a unique hostname.
   perfProfileForMDM	Optional performance profile to set for MDMs: High (default) or Compact. When this field is left empty, the default option is applied. High performance is required for the Fine Granularity data layout feature.
   Virtual IPs	A virtual IP address (VIP) for each possible manager MDM. This virtual IP address can be used for communication between the MDM cluster and SDCs on the PowerFlex data networks (not applicable for the management network). Only one virtual IP address can be mapped to each NIC, with a maximum of four virtual IP addresses per system.
   Virtual IP NICs	The NIC to which the virtual IP addresses are mapped. Maximum length of 11 characters.
   Is SDS	Deploy the SDS component on this node: Yes or No	Yes
   SDS Name	The name for the SDS node, up to 31 characters in length. The name must not contain spaces. Ensure that each server on which SDS is installed has a unique hostname.
   SDS All IPs	SDS IP addresses to be used for communication among all nodes. Maximum of eight addresses, comma-separated, no spaces. NOTE: When using this option, all roles use the same IP addresses, and there is no need to fill the SDS\SDC only columns.
   SDS-SDS Only IPs	SDS IP addresses to be used for communication among SDS nodes. When the replication feature is used, these addresses are also used for SDS-SDR communication. Maximum of eight addresses, comma-separated, no spaces.
   SDS-SDC Only IPs	SDS IP addresses to be used for communication among SDS and SDC nodes only. Maximum of eight addresses, comma-separated, no spaces. For SDC-only nodes, enter the IP address in this column.
   Protection Domain	The Protection Domain to which to assign this SDS or SDR. The name can be up to 31 characters in length.
   Fault Set	The Fault Set to which to assign this SDS. NOTE: When using Fault Sets, you must assign a minimum of three, and follow the guidelines described in “Fault Sets” in the Architecture section of Getting to Know PowerFlex.
   SDS Storage Device List	Storage devices to be added to an SDS. For more than one device, use a comma-separated list, with no spaces. Ensure that the device names and paths do not contain any special characters, such as: ?, `, !, @, # Ensure that devices are prepared as described in "Configuring direct attached storage" in the Architecture section of Getting to Know PowerFlex. Device name format on Linux: /dev/sdb,/dev/sdc When specifying the SDS device path on a Linux node, use the path according to how it is listed in cat/proc/partitions (and not according to the output of fdisk -l). Use persistent device names. For example: fdisk output: /dev/mapper/samah-lv1, /dev/sdb cat /proc/partitions output: dm-3, sdb Use these values in the CSV file: /dev/dm-3, /dev/sdb To enable volume creation, you must add (at least) one device to (at least) three SDSs, where each SDS is in a separate Fault Set, and each device has a minimum of 130 GB free storage capacity (an SDS which has not been assigned to a Fault Set is treated as a Fault Set of its own). You can do that via the CSV file, or at a later stage. The maximum number of devices per SDS is listed in the “Product limits” table in the Getting to Know PowerFlex guide. Device data is erased when devices are added to an SDS. When adding a device to an SDS, PowerFlex will check that the device is clear before adding it. An error will be returned, per device, if it is found not to be clear. PowerFlex might not perform optimally if there are large differences between the sizes of the devices in the Storage Pool—for example, if one device is as big as the rest of the devices. After adding devices, you can define how much of the device capacity is available to PowerFlex by using the SCLI command modify_sds_device_capacity. For optimal performance, try to balance the number of devices of a Storage Pool, and the capacity of those devices among the relevant SDSs. When adding devices that were used in a previous PowerFlex system, follow the instructions in "Prepare previously used SDS devices".
   StoragePool List	Sets Storage Pool names, up to 31 characters in length. Use this option in one of the following ways to assign Storage Pools to the devices in the SDS Storage Device List: A comma-separated list of StoragePool names; the length of the list must be the same length as the list of devices in the SDS Storage Device Names list. In this case, the Storage Pools will be mapped to the devices in that list, respectively. The list must be comma-separated, with no spaces. If one Storage Pool is entered here, the same Storage Pool will be mapped to all the devices in the SDS Storage Device List. If no Storage Pool is listed here, one will be automatically created during installation, named default.
   SDS Storage Device Names	Sets names for devices. Assigns device names to the devices in the SDS Storage Device List, respectively. The amount of names must equal the amount of devices. Comma-separated, no spaces.
   SDS Storage Device Path	The length of the SDS device path must not exceed 63 characters. For example, this path for the following NVMe disk under /dev/disk/by-id is not supported because it is too long: /dev/disk/by-id/Dell_Express_Flash_NVMe_PM1725_1.6TB_SFF_______S2JPNA0J500141 Alternatively, this equivalent name is supported: /dev/disk/by-id/nvme-eui.002538957100082e
   perfProfileForSDS	Optional performance profile to set for SDSs: High (default) or Compact When this field is left empty, the default option is applied. High performance is required for the Fine Granularity data layout feature. For more information, see "Configure performance profile during deployment".
   Is SDC	Yes or No.	Yes
   perfProfileForSDC	Optional performance profile to set for SDCs: High (default) or Compact. When this field is left empty, the default option is applied. High performance is required for the Fine Granularity data layout feature. For more information, see "Configure performance profile during deployment".
   SDC Name	Sets a name for the SDC, which can be up to 31 characters in length. Ensure that each server on which SDC is installed has a unique hostname.
   RFcache	Install the RFcachexcache package on an SDS: Yes or No (default). NOTE: If you wish to configure RFcache devices as part of the deployment, set this column to Yes, RFcache SSD Device List column to a list of comma-separated devices, and the RFcache Accelerated StoragePool List to a corresponding comma-separated list of Storage Pools. NOTE: Ensure that the RFcache feature is supported by the Linux Kernel version used in your system. For more information, refer to the Getting to Know Dell PowerFlex Guide.
   RFcache SSD Device List	List of SSD devices to provide RFcache acceleration for Medium Granularity data layout Storage Pools. Up to eight devices, comma-separated, with no spaces. Ensure that the device names and paths do not contain any special characters, such as: ?, `, !, @, # If RFcache is Yes, and this field is left blank, you can add RFcache devices after installation. NOTE: The RFcache feature is only supported by 512 Byte devices for the cache use case. RFcache is not supported on 4K devices used for cache. However,RFcache acceleration of 4K storage devices is supported.
   RFcache Accelerated StoragePool List	The names of the Storage Pools for which RFcache acceleration is to be provided. The amount of StoragePool names must equal the amount of devices in the RFcache SSD Device List. Comma-separated, no spaces. The specified Storage Pools must be defined in StoragePool List.
   NVDIMM Acceleration Device List	List of NVDIMM devices used to provide acceleration for Storage Pools using Fine Granularity data layout. Up to six devices, comma-separated, with no spaces. Ensure that the NVDIMM devices are configured for DAX mode.
   NVDIMM Acceleration Pool List	Sets NVDIMM Acceleration Pool names. Assigns NVDIMM Acceleration Pools to the devices in the NVDIMM Acceleration Device List, respectively. The amount of NVDIMM Acceleration Pools must equal the amount of devices. Comma-separated, no spaces.
   Is SDR	Deploy the SDR component on this node: Yes or No (default). A minimum of two SDRs per peer system must be deployed for the replication feature. SDRs must be deployed on nodes that also contain the SDS component.
   SDR Name	The name of the SDR node, up to 31 characters in length. The name must not contain spaces. Default: empty field.
   SDR Port	The port used by the SDR. Default: 11088
   SDR Application IPs	The IP addresses through which the SDC communicates with the SDR. When an application writes data, it will work with the SDR through these IP addresses. Maximum of eight addresses, comma-separated, no spaces. The same IP addresses can be used for Application, Storage, External. IPv6 is supported.
   SDR Storage IPs	The IP addresses through which the SDR communicates with the MDM for server side control communications. Maximum of eight addresses, comma-separated, no spaces. The same IP addresses can be used for Application, Storage, External. IPv6 is supported.
   SDR External IPs	The IP addresses through which the SDR communicates with peer systems' SDRs. Maximum of eight addresses, comma-separated, no spaces. The same IP addresses can be used for Application, Storage, External. IPv6 is supported.
   SDR All IPs	SDR IP addresses to be used for communication among all nodes (including all three roles) Maximum of eight addresses, comma-separated, no spaces. This is the default setting.
   perfProfileForSDR	Optional performance profile to set for SDRs: High (default) or Compact. When this field is left empty, the default option is applied. High performance is required for the Fine Granularity data layout feature. For more information, see "Configure performance profile during deployment".
   Is Vasa Provider	A VASA provider is used (for VMware vSphere Virtual Volumes (vVols) support): Yes means that the node contains a VASA provider. Any other value means that it does not contain one. A cluster may contain a single VASA provider, three VASA providers, or none. VASA providers must be installed on nodes where SDC is also installed. NOTE: You can only add VASA providers to a system that previously contained no VASA providers. You must begin from a state where no VASA providers are installed in the cluster.
   Vasa Provider IP	The IP address of the VASA provider NOTE: If no IP address is defined in this column, the IP address will be taken from the "IPs" column.
   Storage Pool Configuration section	Optional section of the CSV file that lets you configure properties for each of the Storage Pools to which the SDS devices are assigned
   ProtectionDomain	The Protection Domain to which the Storage Pool belongs
   StoragePool	The Storage Pool name
   Media Type	The expected device media type in the Storage Pool: HDD or SDD
   External Acceleration	External acceleration type, if used: Read—All devices are accelerated by a non-PowerFlex read cache Write—All devices are accelerated by a non-PowerFlex write cache ReadAndWrite—All devices are accelerated by both non-PowerFlex read and write cache
   Data Layout	Data layout type: FG—Fine Granularity layout MG—Medium Granularity layout
   Fine Granularity ACCP	The name of the Fine Granularity Acceleration Pool associated with the Storage Pool
   Zero Padding	Zero padding enabled on the Storage Pool (mandatory for Fine Granularity): Yes—zero padding enabled No—zero padding disabled
   Compression Method	The compression method used in the Fine Granularity Storage Pool. Compression method might affect performance. For more information about recommended use cases, refer to the Getting to Know Dell PowerFlex Guide. None—no compression is used Normal—compression is enabled
   Replication journal capacity percentage	The percentage of capacity dedicated to the replication journal: a value between 1–100 The percentage is calculated from the net capacity, and not from the total capacity. Use the information provided in the Getting to Know Dell PowerFlex Guide to calculate the percentage of capacity that is required. This calculation also applies to expansion use cases where SDR is being added to an existing system. Each Protection Domain that contains SDRs should have at least one Storage Pool with replication journal capacity higher than 0%. Each SDR needs at least 108 GB of replication journal capacity in order to work. Take this into consideration when calculating capacity requirements.

3. Save the CSV file and retain for future use.