The basic idea behind having multiple subnets (or multiple vlans) for a connection is to evade a broadcast storms. Broadcast storm on a subnet does not go beyond that particular vlan or subnet.
So, in case we had an active connection through one vlan (which suffers a broadcast storm), we can make use the connection through the other vlan.
The VNXe HA document discribes the configuration for multiple levels of redundancy i.e nics on host, Switches connecting and eth ports for iSCSI server.
The eth2 ports on both SP's are connected to switch1 and that of eth3 on switch 2 and hence multiple subnets for eth2 and eth3 for a iSCSI server.
If you have a single switch between VNXe and host you can have single subnet configured for the iSCSI server.
Thanks for your reply - that doesnt clarify anything for me though.
The deisgn I am thinking of would have multiple switches, Vmware hosts and Storage Processors. But I don't understand the need for two iSCSI subnets - or to put another way, would it all work perfectly if only one iSCSI subnet was used?
We also have the option to uplink the two switches if we want, though the whitepaper doesn't show any switch interconnect (no sure if this makes any difference)
The two different subnets are suggested to have redundancy in all levels -
1. On storage you have two SPs - SP hardware redundancy
2. Two NIC used on the SP (iSCSI server) - port redundancy
3. Two switches - for switch level redundancy (datalink Layer)
4. Two subnets - for network (router level) redundancy (Network Layer)
5. Two NICs on host - host level port/NIC redundancy
The subnet level redundancy is not mandatory though. If you already have two subnets in your environment you can utilze them. If you dont, you can always use the same subnet for both interfaces and it works well.
Shardul1
33 Posts
0
May 13th, 2013 22:00
The basic idea behind having multiple subnets (or multiple vlans) for a connection is to evade a broadcast storms. Broadcast storm on a subnet does not go beyond that particular vlan or subnet.
So, in case we had an active connection through one vlan (which suffers a broadcast storm), we can make use the connection through the other vlan.
Sriky
45 Posts
0
May 6th, 2013 08:00
Hello,
The VNXe HA document discribes the configuration for multiple levels of redundancy i.e nics on host, Switches connecting and eth ports for iSCSI server.
The eth2 ports on both SP's are connected to switch1 and that of eth3 on switch 2 and hence multiple subnets for eth2 and eth3 for a iSCSI server.
If you have a single switch between VNXe and host you can have single subnet configured for the iSCSI server.
Regards,
Sri
MagicHair
9 Posts
0
May 6th, 2013 13:00
Thanks for your reply - that doesnt clarify anything for me though.
The deisgn I am thinking of would have multiple switches, Vmware hosts and Storage Processors. But I don't understand the need for two iSCSI subnets - or to put another way, would it all work perfectly if only one iSCSI subnet was used?
We also have the option to uplink the two switches if we want, though the whitepaper doesn't show any switch interconnect (no sure if this makes any difference)
DynaDin
138 Posts
1
May 7th, 2013 00:00
The two different subnets are suggested to have redundancy in all levels -
1. On storage you have two SPs - SP hardware redundancy
2. Two NIC used on the SP (iSCSI server) - port redundancy
3. Two switches - for switch level redundancy (datalink Layer)
4. Two subnets - for network (router level) redundancy (Network Layer)
5. Two NICs on host - host level port/NIC redundancy
The subnet level redundancy is not mandatory though. If you already have two subnets in your environment you can utilze them. If you dont, you can always use the same subnet for both interfaces and it works well.
MagicHair
9 Posts
0
May 13th, 2013 23:00
Thanks all, I think I got it now