topic Re: Network redundancy with HF40 2x2 10GbE ports in Array Setup and Networking

Network redundancy with HF40 2x2 10GbE ports

ixothym1 — Mon, 08 Jul 2019 15:45:49 GMT

Hi all,

new Nimble customer here looking for some advice regarding network setup for our new HF40 array with 2x2 10GbE ports. Currently we deployed a first network setup like this:

Management subnet: eth0a

iSCSI subnet: eth0b

As you can see, each subnet only has one uplink per controller, so pulling a cable or restarting the switch where the active controller is uplinked leads to immediate controller failover.

We're wondering if it's possible to setup the management subnet and the iSCSI subnet as tagged VLANs that are both carried over eth0a and eth0b, as that promises redundancy. Also it seems like a big waste to only run the management traffic (no group setup here) over a 10GbE link

Regards,

Peter

Re: Network redundancy with HF40 2x2 10GbE ports

Nick_Dyer — Mon, 08 Jul 2019 16:10:39 GMT

Hi!,

You definitely need dual paths per network, per controller - otherwise you're architecting for multiple Single Points of Failure.

You have onboard ports to use as well (2 per controller, and are 10Gb/1Gb); typically users assign the management traffic to those ports, bonded across a pair of mgmt switches. Is that something you could look to do?

Alternatively, yes it's entirely possible to create VLANs for both management and iSCSI data subnets to reside on the same physical ports.

Take a look at Infosight, there are a few networking guides there, as well as a hardware guide which details this.

Re: Network redundancy with HF40 2x2 10GbE ports

ixothym1 — Mon, 08 Jul 2019 16:25:53 GMT

Thanks, that's reassuring. We're going to test it tomorrow and post how it worked out!

Re: Network redundancy with HF40 2x2 10GbE ports

giladzzz — Tue, 09 Jul 2019 09:14:12 GMT

you can share management and data ports

below is an output from a Nimble HF20

Nimble OS $ netconfig --info active
Group Management IP: 172.30.10.50
Group leader array: XXXXXXX
Member array(s): XXXXXXX
ISCSI Automatic connection method: Yes
ISCSI Connection rebalancing : Yes

Routes:
---------------+---------------+---------------
Destination Netmask Gateway
---------------+---------------+---------------
0.0.0.0 0.0.0.0 172.30.10.1

Subnets:
------------------------+------------------+---------+---------------+----+-----
Label Network Type Discovery IP VLAN MTU
------------------------+------------------+---------+---------------+----+-----
ISCSI 172.24.0.0/24 Data 172.24.0.9 100 1500
MGMT 172.30.10.0/24 Mgmt 172.30.10.50 10 1500

Array Network Configuration: XXXXXX
Controller A IP: 172.30.10.51
Controller B IP: 172.30.10.52
---------+------------------------+---------------+------
NIC Subnet Label Data IP Address Tagged
---------+------------------------+---------------+------
eth0a.100 ISCSI 172.24.0.24 Yes
eth0a.10 MGMT N/A Yes
eth0b.100 ISCSI 172.24.0.14 Yes
eth0b.10 MGMT N/A Yes

Regards

give a KUDO if this helps

Re: Network redundancy with HF40 2x2 10GbE ports

ixothym1 — Tue, 09 Jul 2019 13:25:11 GMT

Thanks @giladzzz and @Nick_Dyer , it worked flawlessly