• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

EqualLogic PS6210X Network Configuration?

K

KapsZ28

2[H]4U
Joined
May 29, 2009
Messages
2,114
In the past I have configured a PowerVault SAN using two separate subnets and VLANs. This was Dell's recommended setup. I am guessing this is for the multiple paths and load balancing to work correctly. I also setup two 10Gb switches stacked with LAG. On the ESXi host, it is a single Standard vSwitch with two VMkernels. Both on a different subnet. (10.10.10.15 and 10.10.11.16). In NIC Teaming both are set to Active with IP Hash.

Would this be the same configuration for an EqualLogic PS6210X? I am reviewing someone else's work and they have eth0 as 10.0.0.10 and eth1 as 10.0.0.11. I would think they should also be on different subnets.
 
You don't want to do teaming for iSCSI, ever. There's 2 common way's to set that iSCSI multipathing up.

1) 2 or more vKernels in the same vSwitch with each isolated to a different NIC. i.e. Override the failover for all other NICs assigned to that vKernel to unused. This forces each vKernel to only use a single NIC in that vSwitch.

2) Use 2 or more separate vSwitches to isolate the NICs.

On the Equallogic side, all interfaces are typically separate IPs on the same subnet, split between 2 physical switches for redundancy. The EQL controllers will bounce traffic between the IPs associated to the different interfaces to load balance. When you add your iSCSI connection to the EQL, add the GroupIP, not the individual interface IPs.

What switches are you using? How are they setup?
 
Last edited:
Well yeah, it is setup as option 1. One vSwitch with two VMkernels, each with only one active adapter. Not sure why I said both were active. I guess with only one active, IP Hash doesn't really do anything, huh?

So, if that is the case, then I guess there is no point in using LAG for the iSCSI ports on the switches?

I guess you also want the Path Selection set to Round Robin rather than Fixed?

These are two Dell N4032 switches by the way. They are stacked together, but that is about it. LAG is not currently setup. Each ESXi host has four 1 Gb NICs going to different switches, not the N4032. The hosts also have four 10 Gb NICs. Two for iSCSI and two for vMotion.

How about vMotion? There are two additional 10 GbE NICs setup for just vMotion. Any point in setting those up in LAG and using IP Hash? I also read that with a MTU set to 9000, it could increase the time it takes to perform a vMotion when compared to 1500.
 
Multipathing is used instead of LAG.
Either Round Robin or Equallogic has their own customized MPIO plugin for vSphere. Also, the default for Round Robin, is to send 1000iops over a path before bouncing to another one. Accepted performance/stability gains can be had by lowering this number, Googling it, you'll see lots of references to iops=1, which is great for tiny installations, but can be limiting as demand on the array grows. I've seen/used iops=3, works quite well with EQL arrays when lots of VMs are involved. Here's a quick, concise article going a little deeper into that.

Regarding vMotion, What kind of VM density are you talking about and/or how much RAM is in each host. I've rarely seen an environment that needs more than a single 10GbE NIC for vMotion.
 
Last edited:
Ah yes. I remember changing the IOPS on one of our setups.

This is a customer's network that I am looking into. Unfortunately they didn't label the ports or provide a diagram of which host is plugged into which switch. Hopefully each host is plugged into a separate switch. I guess I will have to look for the MAC addresses to confirm.
 
With only have a single EqualLogic PS6210X SAN, would you recommend using RAID 6, 10, or 50?
 
Traditional RAID differences still apply here:
RAID 10 will be the fastest, but offer the least usable space.
RAID 50 will offer better capacity, and still reasonable performance, but loses a bit in redundancy
RAID 6 is the most redundant, but by far the slowest performer.

What's your balance between Performance, Capacity and redundancy? I've been doing RAID50 in EQLs for a lot of years, and until last year I hadn't seen an unrecoverable failure directly attributed to drive failure that wasn't cause by something else first. But that could be luck of the draw too. Without knowing more about the load to be placed on it, I'd say RAID6 unless you have an express need for more performance. Being a 24x2.5" chassis, minus the 2 HDD's used as global spares, You are effectively Doing RAID among 22 HDDs and that's a decent # of spindles to share to load, so unless you are doing something with crazy random access like VDI, you'll probably be fine with RAID6.
 
You must log in or register to reply here.
Back
Top