NS0-157 · Question #31
You notice your indirect I/O performance between nodes in your 4-node cluster is slowing down from prior week. Each node has two 10 GB connections to the cluster interconnect, one to each switch…
The correct answer is B. Add two more cluster interconnect connections per node. When cluster interconnect utilization exceeds 80% and unused switch ports are available, adding more per-node connections is the correct way to increase total bandwidth.
Question
You notice your indirect I/O performance between nodes in your 4-node cluster is slowing down from prior week. Each node has two 10 GB connections to the cluster interconnect, one to each switch. You are using four out of 12 available ports on each cluster interconnect switch. Your cluster network utilization is over 80%. What should you do to increase cluster interconnect performance?
Options
- AChange the SFP+ from 10 Gb to 16 Gb on each node and switch.
- BAdd two more cluster interconnect connections per node.
- CAdd additional ISLs between the two cluster interconnect switches.
- DChange the cable from TwinAx to optical with SFP+ for better throughput.
How the community answered
(23 responses)- A13% (3)
- B52% (12)
- C30% (7)
- D4% (1)
Why each option
When cluster interconnect utilization exceeds 80% and unused switch ports are available, adding more per-node connections is the correct way to increase total bandwidth.
16 Gb SFP+ is a Fibre Channel speed designation and is not applicable to Ethernet-based cluster interconnects; this is not a valid upgrade path for 10GbE cluster ports.
Each node currently has two 10 Gb uplinks providing 20 Gb of cluster interconnect bandwidth. Adding two more connections per node increases each node's aggregate bandwidth and improves load balancing for indirect I/O, directly addressing the utilization bottleneck. The existing switches already have 8 unused ports per switch to support this expansion without additional hardware purchases.
Adding ISLs between the two cluster interconnect switches improves inter-switch bandwidth but does not increase the per-node uplink bandwidth that is causing the high utilization.
Switching from TwinAx to optical fiber does not change the 10 Gb port speed or throughput; both are standard 10GbE media types with equivalent bandwidth.
Concept tested: ONTAP cluster interconnect bandwidth scaling with additional links
Source: https://docs.netapp.com/us-en/ontap/networking/display_cluster_connections.html
Topics
Community Discussion
No community discussion yet for this question.