Storage Boards Cleanup
To make it easier to find information about HPE Storage products and solutions, we are doing spring cleaning. This includes consolidation of some older boards, and a simpler structure that more accurately reflects how people use HPE Storage.
Disk Arrays
cancel
Showing results for 
Search instead for 
Did you mean: 

EVA 8100 cabling question

itchic
Occasional Visitor

EVA 8100 cabling question

Scenario two MDS switches, ISL'd and multiple vsans each can see. Servers with single hba's. Storage is eva8100. If an MDS switch goes down, how can the single pathed systems stay up with out reconfiguring zoning and vsans? Can eva controller pairs be split between MDS switches and vsans as follows, ports 1 & 2 to one mds switch, each on different vsans then same for ports 3 & 4. Is this possible and if not, why?
MDS1-------ISL---------MDS2
vsanA --1 8100A 3--vsanB
vsanA --2 8100A 4--vsanB

This doesn't work with out causing an outage, requiring reconfiguration:
MDS1-------ISL---------MDS2
vsanA --1 8100A 2--vsanB
vsanA --3 8100A 4--vsanB

3 REPLIES
itchic
Occasional Visitor

Re: EVA 8100 cabling question

I wrote this wrong, I meant can I do this:
MDS1-------ISL---------MDS2
vsanA --1 8100A 3--vsanA
vsanB --2 8100A 4--vsanB
I want to split the pairs between MDS switches but keep vsans and zoneing the same.
Bulent ILIMAN
Trusted Contributor

Re: EVA 8100 cabling question

You cannot guarantee to stay up a single HBA server what ever you do. If the switch that HBA connected goes down then you are down. Single HBA itself is a Single Point of Failure(SPoF) on the other hand.

EVA cabling should be done in the following rule:
Odd numbered ports of both controllers to one switch and Even numbered ports of both controllers to the other switch. Usually we dont connect the switches to each other. It only duplicates the number of paths but brings no redundancy or performance.

You need second HBA's to your hosts to have actually no SPoF.
itchic
Occasional Visitor

Re: EVA 8100 cabling question

I have the switches ISL'd so I can bring one down for an upgrade with out downtime. I have had failures also, and the configuration most certainly keep us up. The vsans, fabric is seen from both, they are redundant. I can litterally turn one off and run with one exception. Right now the odd ports are connected to one MDS switch on the same vsan. The server HBA is zoned for them. When that MDS when down, then that server couldn't see storage. If I loose port 1 on a controller, then 3 is still up. I got them impression they were a failover pair so if they (1,3) are connected to the same switch that defeats the purpose.
That's what prompted my question. If they are a pair, evens and odds can they be physically connected to two different switches? Just like if the server had dual hba's. Each pair would be in the same vsan, zone etc.