This section covers step 2 defined in the Introduction section.
Every predefined model defines machine_type attribute for boot disk and a minimum of two disks (one set) for storage (referred to as initial disks or default storage disks or default disks). The default storage disks for decoder service are named (disk_name attribute in host-models.yml) as decoder and decodersmall. The default disks for concentrator are index and concentrator. The disk naming convention is also tied to subsequent configuration steps such a creation of partitions, volume allocation to service. It also helps to associate the corresponding partition scheme that is created on that particular disk when using NW REST API to complete storage configuration
Each node (Decoder or Concentrator) has its own storage attributes defined under disks (Refer to screen shot below). The attributes are further classified into two groups for each service with each group corresponding to a specific logical volume name. For decoder service, the two groupings refer to the logical volume names: decoder and decodersmall and for concentrator service these are concentrator and index.
Note: The grouping names (decoder and decodersmall for decoder and concentrator and index for concentrator) under disks match the disk_name attribute values that they define. This relationship (underlined in red below) MUST be maintained when defining a Custom storage model since the disk creation process requires this 1-1 mapping of grouping name to disk_name.
host-models.yml snippet highlighting the disk groupings and disk_name relationship (underlined in red):
Each disk attached to a SASE Decoder node is configured as a separate logical volume to host the service’s database(s) such as packetdb, metadb, sessiondb and index. The disk named (disk_name attribute in host-models file) as decoder corresponds to decoder volume and hosts packetdb and the disk named (disk_name attribute in host-models file) as decodersmall corresponds to decodersmall volume and host sessiondb, metadb and index. If decoder storage requirements mandate larger disks (i.e 12 days of Packet retention versus 6 days of packet retention or 23 days of packet retention versus 12 days or 6 days) then multiple sets of disks are defined and created.
If a model defines two sets of disks, the first is the default disks named (disk_name) as decoder and decodersmall and the second set is named as decoder0 and decodersmall0, likewise when a model defines four sets of disks ex: c1r23m120, first set is default disks (named decoder and decodersmall) , second set of disk named as decoder0 and decodersmall0 and the third set of disks are named as decoder1 and decodersmall1 and fourth set is decoder2 and decodersmall2. Any additional disk set follow the similar naming convention as described above. i.e decoder<COUNTER> and decodersmall<COUNTER> where COUNTER is incremented by 1 for any new set of disks starting with 0. The first part of the name (decoder , decodersmall) helps in associating the disk with volume name/type that is created later. The volume name is used to identify the appropriate Decoder service’s database during service allocation.
Similar pattern is followed with concentrator disks. The default disk set is index and concentrator. The concentrator is pd-standard (disk_type) and index is disks are pd-balanced.The index disk is used to host index database and concentrator disk is used to host root, sessiondb and metadb. When more than one disk set is created, these are named as index<COUNTER> and concentrator<COUNTER> where COUNTER starts with 0 and increments by 1 (similar pattern as decoder disks described above). Refer to above screen shots for pattern identification for c1r12m60.
The additional_storage attribute value in sase-deployment-models.yml determines whether the storage disks are created or not. A value of true creates the disks and false skips the disk creation. The default value is false. Updating the additional_storage to true and re-running the nw-create-cloud-hybrid creates the storage disks. These disks are not deleted by re-running nw-create-cloud-hybrid after setting additional_storage to false.
The storage requirements must be identified to select the appropriate storage model. The retention days along with capture rate is used to identify the appropriate model. After the model is identified, the model_name value for every SASE node must be updated in the sase-deployment-models.yml file. If the multiple nodes are deployed, then each node’s model_name must be assigned with appropriate value.
For both predefined and custom models, after identifying the model (Refer to Identify a Pre-defined Storage Model or Define a Custom Storage Model section(s) below) the model_name attribute value (in sase-deployment-models.yml) is set to the identified model name.
Refer Appendix B for SASE Decoder Storage Configuration and Appendix C for SASE Concentrator Storage Configuration.