The following table provides requirements for the number of servers needed to support any number of devices in production. The specifications below have been verified up to a scale of 9 million modems with a total of 200 million SCQAM channels (combined upstream and downstream counts) and 9 million OFDM/A channels across all devices.
| Type | Qty | vCPU | Memory | Disks |
|---|---|---|---|---|
| Data Node | n1 | 8 | 28 GB | 1 x 30GB OS 1 x 40GB /opt/arris 2TB SSD Data 1 x 32GB SSD Commit Log |
Data nodes must be deployed in groups of 3, across 3 racks. The node quantity must be 3, 6, 9, and so on, with a minimum of 3. Designed to scale out horizontally and operate in an active/active operational mode. For more details, or if your history is longer or shorter, refer to Requirements for data node servers. Each Cassandra node with 2TB data disk can support up to 10 million channels with 14 days of history in high availability. Default paths:
| ||||
| Cache Node | n+1 | 8 | 32 GB | 1 x 30GB OS 1 x 40GB /opt/arris |
Designed to scale out horizontally and operate in an active/active operational mode. Each NXT Engine node is capable of supporting 14 million channels in high availability or 10.5 million channels if FieldPro is enabled. To achieve high availability, at minimum one more NXT Engine node should be provisioned above the quantity required to meet the desired scale.2 Billing and topology ingest should be run at a maximum once every two hours. If ingest occurs more frequently, then SOLR node CPU and memory performance can be affected, which will cause the SOLR secondary node to enter recovery mode. | ||||
| Search Node | n | 4 | 8 GB | 1 x 30GB OS 1 x 40GB /opt/arris |
Designed to scale out horizontally and operate in an active/active operational mode. Each Solr node is capable of supporting 34.5 million channels in high availability.2 | ||||
| Application/UI Node | 2 | 4 | 8 GB | 1 x 30GB OS 1 x 40GB /opt/arris |
| Proxy Node | 1 | 2 | 2 GB | 1 x 30GB OS 1 x 40GB /opt/arris |
| One Proxy node can be used. If more than one
Proxy Node is provided, they can be used in High Availability mode. Note:CommScope recommends that customers use their own hardware-based load balancer, and only utilize a Proxy Node if no other load balancing option is available. | ||||
| Trap Receiver Node | 1 or 2 See notes | 4 | 4GB | 1 x 30GB OS 1 x 40GB /opt/arris |
| Note:
Trap Receiver nodes cannot reside behind a load balancer. One (1) Trap Receiver server can handle up to 5 thousand trap messages per second with a drop rate lower than 5%. Two (2) Trap receiver servers can be used in high availability mode. When utilizing a second Trap Receiver Node, a virtual IP must also be installed for the SNMP Traps to be sent to. Keepalive is installed and configures which Trap Receiver is Primary, and will fail over to the backup node if the primary goes out of service for any reason. This does not increase the overall number of traps that can be managed. | ||||