From the Nutanix perspective, the only thing that is required to connect with the FlashArray is to configure a dedicated storage network. All of the multipathing and optimizations will automatically be handled and configured at the Nutanix level. There is nothing that needs to be configured or changed by the user after getting the storage networking configured.
AHV Network
An AHV network is a VLAN-backed virtual network in Nutanix Acropolis Hypervisor (AHV). Use the information in this section to configure an AHV network.
You should configure NIC interfaces ito distribute NVMe/TCP data storage traffic in different VLANs, enable VLAN tagging, and NIC Bonding as recommended.
Although 1500 bytes is still supported, we recommend that you configure interfaces with MTU 9000 bytes.
| Virtual Switch | Physical NIC | Network Traffic Type | Supported Bond Modes |
|---|---|---|---|
|
vs0 |
Eth0 Eth1 |
Nutanix Management Live migration DR replication Application traffic |
Active/Active or Active/Backup |
|
vs1 |
Eth2 Eth3 |
NVMe/TCP storage traffic |
Active/Active with mac pinning |
|
N/A |
Server IPMI |
Out-of-band management |
N/A |
Multipathing
The multipathing configuration is handled completely with NCP. There is nothing that needs to be configured by the user for multipathing. Nutanix will be leveraging Linux native multipathing, configure udev rules and path selection policy automatically.
Timer Settings
The timer settings listed in this section are provided for information only. These values are automatically configured when Nutanix Cloud Platform (NCP) connects to the FlashArray. Customers are neither required nor permitted to modify these timers.
| Parameter | Recommendation |
|---|---|
|
Keep Alive Timeout (KATO) |
5 sec |
|
Controller Loss Timeout |
3600 sec (60 mins) |
|
NTNX CQT (Command Quiesce Time) |
30 sec |
Compute Nodes Performance Tuning
You should review the node-level settings in this section in collaboration with the Nutanix design and support teams. This section is for informational purposes as it will be handled automatically
Host and guest networking configurations can have a significant impact on end-to-end performance, particularly for high-throughput and latency-sensitive workloads. See additional guidance and references in the Resources section of this guide.
| Parameter | Description | Recommendation |
|---|---|---|
|
Receive Side Scaling (RSS) |
Distribute RX load across CPU cores |
Enable |
|
Generic Receive Offload (GRO), Large Receive Offload (LRO), and Transmit Segment Offload (TSO) |
Offload segmentation/reassembly |
Enable if supported |
|
Interrupt Coalescing |
Reduce CPU overhead |
Tune |
|
NIC Ring Buffers |
Prevent packet drops under burst |
Maximize RX/TX buffers |
|
Queue Depth |
Storage I/O concurrency |
Use default from Pure multipath recommendations |