Configuring two PowerConnect 5424 Switches for ISCSI Backbone.

Question

Configuring two PowerConnect 5424 Switches for ISCSI Backbone.

We had a vendor come in and setup our VMware environment with a SAN. For the ISCSI backbone, they basically just plugged in the 2 PowerConnect 5424 switches stacked and left everything else default. Recently we have been having major latency issues between the VM hosts and our SAN. VMware support says everything looks fine from their end. They even ran a tool called esxtop with the u option to show me. They said more than likely the SAN is overloaded. So we contacted our SAN Support and they say everything appears to be fine. The arrays don’t appear to be overloaded. This is when I began digging into the switch configuration to find out that no configuration was ever done to optimize ISCSI. I plan on configuring the switches on a weekend but would like to run this past some experts to make sure I am not missing anything on the switch side. We have two Power Connect 5424 switches connected through G24. I plan on setting up two ports on the switches as LAG ports. STP is currently turned on, with no flow control, and ISCSI is disabled as well. Also I have seen were you would definitely want rstp on vs stp. One more thing I should add is that our vm hosts have 4 nics each. 2 nics are plugged into switch 1 and the other 2 into switch 2. Our SAN has 4 nics as well and they are split up the same way. Looking at the counters, there is no type of load balancing going on. Only one nic is being used regardless for the vm hosts and the san as well. I have checked the counters on the switches and havent seen any packet loss.
Someone in another forum recommended creating another vlan and putting all the ports that are being used by hosts and the SAN in a different vlan besides the default vlan 1

Maybe a Vlan 10 on both switches

console(config)# vlan database
console(config-vlan)# vlan 10
console(config-vlan)#name iscsi
console(config-vlan)#ip address 172.16.0.10 255.255.255.0 (Different IP for Switch 2)
console(config-vlan)# exit
console(config)# interface range g1 – 22
console(config)# switchport mode access vlan 10

SETUP LAG on both switches

cosole> enable
console# configure
console (config)# interface range ethernet g23,g24
console(config-if)# channel-group 1 mode on
console(config-if)# exit
console(config)# interface port-channel 1
console(config-if)# flowcontrol on

Do I need to set the LAG ports as trunks?
Turn Flow control on for all ports on both switches

console> enable
console# configure
console (config)# interface range ethernet all
console(config-if)# flowcontrol on

Enable ISCSI on both switches ( Since these switches are on their own network, will doing this command make a difference. I am under the impression it enables some type of QOS.
console (config)# iscsi enable

Enable Jumbo Frame on both switches
console> enable
console# configure
console(config)# port jumbo-frame

Enable RSTP on all ports (Both switches) being used by vmware hosts and the SAN (not on the LAG Ports)?
console(config)# spanning-tree mode rstp
console(config)# interface range ethernet g1-22
console(config-if)# spanning-tree portfast

Copy running-config startup-config

Should i leave RSTP off on the LAG ports?

Any thoughts on if I am on the right path or are there any changes i should consider before implementing this? Any help would be greatly appreciated. Also please let me know if there is additional information needed.

Thanks,

DELL-Willy M · Answer

Here is some best practice information for ISCI

www.dell.com/.../ip-san-best-practices-en.pdf

Ether Flow Control: Dell recommends that you enable Flow Control on the switch ports that handle iSCSI traffic. In addition, if a server is using a software iSCSI initiator and NIC combination to handle iSCSI traffic, you must also enable Flow Control on the NICs to obtain the performance benefit. On many networks, there can be an imbalance in the network traffic between the devices that send network traffic and the devices that receive the traffic. This is often the case in SAN configurations in which many hosts (initiators) are communicating with storage devices. If senders transmit data simultaneously, they may exceed the throughput capacity of the receiver. When this occurs, the receiver may drop packets, forcing senders to retransmit the data after a delay. Although this will not result in any loss of data, latency will increase because of the retransmissions, and I/O performance will degrade.

Spanning-Tree Protocol: It is recommended that you disable spanning-tree protocol (STP) on the switch ports that connect end nodes (iSCSI initiators and storage array network interfaces). If you still decide to enable STP on those switch ports, then you should turn on the STP FastPort feature on the ports in order to allow immediate transition of the ports into forwarding state. (Note: FastPort immediately transitions the port into STP forwarding mode upon linkup. The port still participates in STP. So if the port is to be a part of the loop, the port eventually transitions into STP blocking mode.)

Note: PowerConnect Switches default to RSTP (Rapid Spanning Tree Protocol) an evolution in STP that provides for faster Spanning tree convergence and is preferable to STP

Note: The use of Spanning-Tree for a single-cable connection between switches or the use of trunking for multiple-cable connections between switches is encouraged.

Here is a post from our forum discussing the same topic:

en.community.dell.com/.../19357247.aspx

Hope this helps,

Keep us updated if you can.

Networking General

Configuring two PowerConnect 5424 Switches for ISCSI Backbone.

Was this post helpful?