I’m planning the connectivity infrastructure, and have the following questions regarding the network configuration:
How are these interfaces named on the operating system for each of the above ports? It appears from a figure in the VNXe High Availability Guide that the ports labeled 2 and 3 are referenced as eth2 and eth3. Are the ports labeled e0, e1, e2, e3 referenced as eth10, eth11, eth12, eth13?
To configure the Management LAN interface. I read that you run the VNXe Connection Utility to assign IP address, subnet, broadcast, etc. I also read that the Management services for VNXe run on one SP at a time. It does not matter which SP it runs on.
My question therefore is, when you run the connection utility does it randomly select the management port from either SPA or SPB and assign the IP address to that interface? What if the SP that owns the IP address fails, how does the interface on the other SP acquire the IP address?
The ports labeled 2 and 3 are referenced in many documents as iSCSI ports. The VNXe3100 System Installation Guide on Page 15 – “Attach the VNXe3100 to your network”, shows cable 3 (SPB Data) and cable 4 (SPA Data) connected to the LAN and going to SPB iSCSI port 2 and SPA iSCSI port 2 respectively. The page also states: Use a CAT5e or better Ethernet cable to connect the RJ45 port labeled 2 to the network from which you will access your data on the VNXe system.
Our plans are to utilize the entire storage on the VNXe for VMware via NFS datastores. We really have no requirements for data access via iSCSI. My questions are, is connection to the iSCSI ports (2 and 3) required as described in the installation guide? Wouldn’t I just physically connect and utilize the 4 ports on the optional Ethernet I/O personality module (ports e0, e1, e2, e3)?
Are ports (2 and 3) strictly iSCSI or can they be used for TCP traffic such as NFS instead? The reason I ask this question is Page 17 and 18 of EMC VNXe High Availability show an example of HA configuration for NFS Storage using an LACP group consisting of eth2 and eth3 ports. For that matter, every example I see in any manual only references ports eth2 and eth3, no reference to ports eth10-eth13.
I would like to use as many data ports as are available to me to aggregate into an LACP group. In EMC VNXe Series Storage Systems A Detailed Review, on page 31, it states “Both models in the VNXe series support link aggregation. With link aggregation, up to four Ethernet ports can be combined into one logical link.” and “VNXe allows you to combine the first port from the on-board I/O ports (or the Flex I/O module) to any other port on the same card.”
Is this implying I can create (2) LACP groups; one group consisting of (2) ports (eth2 and eth3) and the other group consisting of (4) ports (eth10, eth11, eth12, eth13)?
Page 3 of the VNXe3100 System Installation Guide under Network Connections states “Two to four 1-Gigabit Ethernet data connections for CIFS, NFS, and iSCSI connectivity”. If I utilized the (2) iSCSI ports and the (4) ports from the optional Ethernet I/O personality module, shouldn’t this say two to six? Is there some reason I can’t use 6 data connections?
Thanks in advance for any assistence in clarifying these questions for me.
How are these interfaces named on the operating system for each of the above ports? It appears from a figure in the VNXe High Availability Guide that the ports labeled 2 and 3 are referenced as eth2 and eth3. Are the ports labeled e0, e1, e2, e3 referenced as eth10, eth11, eth12, eth13?
thats correct. eth2,3 as base connectivity ports on each SP and eth10-13 on each i/o module.
To configure the Management LAN interface. I read that you run the VNXe Connection Utility to assign IP address, subnet, broadcast, etc. I also read that the Management services for VNXe run on one SP at a time. It does not matter which SP it runs on.
My question therefore is, when you run the connection utility does it randomly select the management port from either SPA or SPB and assign the IP address to that interface? What if the SP that owns the IP address fails, how does the interface on the other SP acquire the IP address?
If the managing SP of the mgmt port fails, the peer SP starts the management stack, as it is described in the network HA white paper. this is not visible to the user. when you initialize your vnxe use a switch where both mgmt ports and your laptop/workstation is connected to, to get a connection to both SPs at the same time during initialization.
The ports labeled 2 and 3 are referenced in many documents as iSCSI ports. The VNXe3100 System Installation Guide on Page 15 – “Attach the VNXe3100 to your network”, shows cable 3 (SPB Data) and cable 4 (SPA Data) connected to the LAN and going to SPB iSCSI port 2 and SPA iSCSI port 2 respectively. The page also states: Use a CAT5e or better Ethernet cable to connect the RJ45 port labeled 2 to the network from which you will access your data on the VNXe system.
Our plans are to utilize the entire storage on the VNXe for VMware via NFS datastores. We really have no requirements for data access via iSCSI. My questions are, is connection to the iSCSI ports (2 and 3) required as described in the installation guide?
there is no dedicated port for any of the usable protocols on your VNXe. you can use any port for any of the protocols (iscsi, cifs, or nfs). or just team them (link aggregation - lacp) together or create a multipath connection using iscsi.
I would like to use as many data ports as are available to me to aggregate into an LACP group. In EMC VNXe Series Storage Systems A Detailed Review, on page 31, it states “Both models in the VNXe series support link aggregation. With link aggregation, up to four Ethernet ports can be combined into one logical link.” and “VNXe allows you to combine the first port from the on-board I/O ports (or the Flex I/O module) to any other port on the same card.”
Is this implying I can create (2) LACP groups; one group consisting of (2) ports (eth2 and eth3) and the other group consisting of (4) ports (eth10, eth11, eth12, eth13)?
exactly. you can group eth3 to eth2 to get a 2 port link and also group 13,12 and 11 to eth10 to get a 4 port link.
Page 3 of the VNXe3100 System Installation Guide under Network Connections states “Two to four 1-Gigabit Ethernet data connections for CIFS, NFS, and iSCSI connectivity”. If I utilized the (2) iSCSI ports and the (4) ports from the optional Ethernet I/O personality module, shouldn’t this say two to six? Is there some reason I can’t use 6 data connections?
no. when you create a shared folder server (for cifs, nfs) you have to choose through which SP the storage will be presented. either SP-A or SP-B. so in this step you decide which lacp trunk you choose. either eth2,3 or eth10,11,12,13 (when trunks desired - otherwise just simply one eth port.) so this is a maximum of 4 ports.
when you create an iscsi server, you also have to choose which SP will present a lun to the host(s). there you have the ability to choose one eth port. after the iscsi server is created you have the possibility to add another network interface port to it (with second IP adress) to create a second path to one lun you are presenting through the iscsi target. here you have a maximum limit of 2 ports to the desired lun.
I was able to get my VNXe3100 on the air today and validated your response "there is no dedicated port for any of the usable protocols on your VNXe. you can use any port for any of the protocols (iscsi, cifs, or nfs)".
Below is a net/port show from my VNXe3100 that shows by default that the Supported interface types for my network interfaces (eth10-13, eth2 and eth3) as being iscsi (Interface for iSCSI storage) and net (Interface for file-based storage, such as shared folders).
Based on that, I would suggest that EMC removes or clarifies the references to ports 2 and 3 as being iSCSI ports on pages 11, 17, 18, 20, 25 and 26 of the "VNXe3100 Hardware Information Guide".
Sebastiank1
3 Posts
1154
0
Posted November 18th, 2011 14:00
Hi Dave,
thats correct. eth2,3 as base connectivity ports on each SP and eth10-13 on each i/o module.
If the managing SP of the mgmt port fails, the peer SP starts the management stack, as it is described in the network HA white paper. this is not visible to the user.
when you initialize your vnxe use a switch where both mgmt ports and your laptop/workstation is connected to, to get a connection to both SPs at the same time during initialization.
there is no dedicated port for any of the usable protocols on your VNXe. you can use any port for any of the protocols (iscsi, cifs, or nfs). or just team them (link aggregation - lacp) together or create a multipath connection using iscsi.
exactly. you can group eth3 to eth2 to get a 2 port link and also group 13,12 and 11 to eth10 to get a 4 port link.
no. when you create a shared folder server (for cifs, nfs) you have to choose through which SP the storage will be presented. either SP-A or SP-B. so in this step you decide which lacp trunk you choose. either eth2,3 or eth10,11,12,13 (when trunks desired - otherwise just simply one eth port.) so this is a maximum of 4 ports.
when you create an iscsi server, you also have to choose which SP will present a lun to the host(s).
there you have the ability to choose one eth port. after the iscsi server is created you have the possibility to add another network interface port to it (with second IP adress) to create a second path to one lun you are presenting through the iscsi target. here you have a maximum limit of 2 ports to the desired lun.
.
kind regards,
Sebastian