notes/projects/neosphere/qumulus/overview-qumulus_and_comp-nodes.md

Infrastucture

5 Node Storage Cluster (Qumulo)

• IT-Glue

• Each node has 240 TB HDD Storage, which in total makes 1.2 Peta Byte of Storage

• Each node has 256 GB SSD Storage for the boot drive

• Each node has 8x 1 TB Nvme SSD Drives (Not sure for what; probably chaching to speed up read and writes which are limited physically by the spinning HDD drives)

• Each node has a dual 25Gbits NIC which are configured in "fault-tolerance"-mode bond (failover backup if one fails) wiht the IP addresses:

  • Primary: 192.168.60.11 - .15
  • Secondary (floating): 192.168.60.21 - .25

• The mgmt dashboard is reachable via each of the IP addresses; e.g. http://192.168.60.13/

• Files are configured using NFS (network file system) and SMB

  • NFS shares to computing cluster (folders /qumulo-vol0000(1/2) on the Qumulo)
  • NFS and SMB shares to client devices

• The configuration for the shares can be modified in the dashboard and are partly configured by the customer and the Qumulo team

3 Node Computing Cluster

• ssh pass for all ubt's

• Each node has a dual 25Gbits NIC which are configured in "fault-tolerance"-mode bond (failover backup if one fails) with the IP addresses:

  • Primary: 192.168.60.200 - .202
  • Secondary (floating): 192.168.60.210 - .212

• Clustered via Software managed by Bjoern Schwalb

• Server ubt-01/02/03 has a NF export of Qumulo node 1/2/3 mounted at /mnt/qumulo-vol0000(1/2)

2x Aruba Switches (SG48LP00G8)

• neo-sw-ug-02

• neo-sw-ug-03

• Each Switch has 16x 25 Gbit/s and 2x 100 Gbit/s Interfaces

• the Switches are coupled via a 100 Gbit/s fiber cable

• Each node of the storage/computing cluster has one of its interfaces of the 25 Gbit/s NICs connected to one switch