Whatever an Intranet is for you - even an iSCSI storage network needs the ability to manage or monitor switches and SAN/NAS, so how else you want to do the job without some router/firewall. In absence, it's a good policy to reserve either the highest or the lowest IP from the IPv4 subnet for a gateway at lest, and configure the address as a default gateway.
Last but not least, I'm not sure how smart it is to use one of these well-known often default LAN IP subnets like 192.168.0.0/24 or 192.168.1.0/24 for such an "Intranet".
Needless to say, a "network" (more appropriately a "broadcast domain") and IPv4 subnet can be created using one or more switches, multiple broadcast domains and IP subnets can be configured on one or multiple swiches using VLANs.
using Microsoft S2D, so actually better to do two different subnets
because setup as one subnet, was getting failoverClustering events about partitioned network when setup as same /23; i.e. both switches set to subnet mask 255.255.254.0 and default gateway 192.168.0.1, where I put the second switch with default 192.168.1.239, and all it's attached nic as 192.168.1.x--could have gone for /24; guessing you're saying creating a vlan was the way to go, but regardless, s2d doesn't want same subnet on the cluster only networks
so moot question now; the first switch setup dhcp disable, left self ip addr as default 192.168.0.139, subnet mask 255.255.255.0, and default gateway 192.168.0.1, and the second switch setup basically the same, but 1.239 self ip, and 1.1 default gateway; no vlan setup (and no failoverClustering events)
my next problem is how to get into knowing port flow control (vs global pause) and QoS, is it RDMA capable; I think I bought a cheap switch, doesn't have much in the way of settings; will have to do testing between servers to confirm traffic is what I want (not really easy conversation in this community; would be more windows powershell community question)
and using well know local/intranet subnet not an issue for me (no one can access it, so not like they can hack it)
and more importantly, thank you for replying (always discouraging when you have a question that no one is willing to even touch!)
using Microsoft S2D, so actually better to do two different subnets
Ways beyond of the Netgear community scope however I'd say.
Show us the layout intended. Indeed there are different subnets required - but i would keep the network management VLAN connected and accessibe direct. This is what the primay IP config is for.
The subnets (VLAN if you want to share the bandwidth) you run for connecting the vNIC for host management (RDP, AD and so on - GbE is sufficient, could be alos the switch management VLAN), plus two vNIC connctions for Storage (S2D, SMB, Live-Migration - I guess this is where you intend to use these two 10G switches).
Last but not least, stay far away using these common IP subnets like 192.168.0.x/24 or 192.168.1.x/24 or 10.0.0.x/24 as these are often default on many simple routers. Choose something meaningful. it's not about hacking or security - it's about logic...
unsure what the best combo is, but I have all nic1 connected to one XS716E switch, and all nic2 connected to the other XS716E, and neither interconnected (and I can change up their subnets, so it's not common 10. or 192.168.)
right now, if I make my 2g nic team (that's on the internet) clusterAndClient, S2D basically ignores the pair of cluster only 10g intranet (just heartbeat maybe goes through, not SMB traffic)
if I make said team none cluster traffic, I get failoverClustering event 1229 (understandably, "Ensure that there is one or more networks that are enabled for cluster use with the 'Allow clients to connect through this network'"), however, then I see SMB traffic on the intranet, and my internet file transfers speed up (as it's not being shared with cluster traffic); so, I'll sometimes switch to this bad config to get some data pushed (but then switch back to proper setup, but smb not hitting intranet is still a problem for me)
any chance you know if the XS716E allows a port to be trunc-ed (not access mode)? or do I just need to interconnect them (setup that last SFP+ to interconnect the two switches)? wondering if I get "ping remote-ipaddr-on-second-switch -S local-ipaddr-frist-switch" to work (agian, haven't dived into vlan yet), would that signal ok to use for cluster smb traffic
and thinking I have to give up RMDA, albeit get-netAdapterRDMA says true, get-smbClientNetworkInterface says not RDMA capable
so for now, I need to go against "recommended" rdma for s2d, and just tinkering with my intranet to see if it can push SMB traffic (like perhaps I need to pretend they have rdma, and after adding QoS via powershell, reboot, my get-smbClientNetworkInterface should show capable?); if I was 100% certain my lom was rdma, I might attempt to follow https://github.com/microsoft/SDN/blob/master/Diagnostics/S2D%20WS2016_ConvergedNIC_Configuration.doc...
this is definitely not specific information, and waay above a casual conversation (I hate being a one man IT shop and in over my head! ugh)
trying to avoid buying "ql41162hlrj" (or switch away from rj45 and going with more options by mellanox or chelsio or intel), and then do I need a new pair of routers (money pitt!)
Model: XS716E|16 Port 10-Gigabit Ethernet Smart Managed Plus Switch with 1 Copper/SFP+ Combo Port
