Home > Storage > PowerFlex > White Papers > MongoDB on Dell PowerFlex with NVMe over TCP > Logical architecture
The following diagram shows the logical architecture of this solution:
Figure 4. Logical design of MongoDB with PowerFlex
PowerFlex Manager (PFxM) is used to deploy a PowerFlex storage-only (SO) cluster using NVMe/TCP template on four bare metal servers followed by the deployment of PowerFlex compute-only (CO) cluster on three PowerFlex nodes. This deployment is fully automated by PFxM.
For the NVMe/TCP configuration, follow the steps that are mentioned in the NVMe/TCP initiator configuration section to configure the initiator on ESXi. At the storage side, the SDT component acts as NVMe target which works along with the SDS component.
To learn more about NVMe/TCP initiator-target communication, see PowerFlex with NVMe over TCP (NVMe/TCP).
A single PowerFlex protection domain and a single storage pool are configured on PowerFlex. The PowerFlex storage pool topology consisting of four SDSs and six SSD disks in each of the four SDS nodes in the PowerFlex system. The volumes are created from the storage pool and mapped to three NVMe hosts (compute-only nodes) where the MongoDB cluster has been built and database software is installed.
After you configure the NVMe/TCP setting on the compute nodes, the MongoDB cluster is built to run the workload tests. YCSB tool is used for performance testing and running on client machines.
For more information about the configuration of MongoDB VMs and the PowerFlex nodes that are used in this solution, see Configuration details.