• Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 1811
  • Last Modified:

All Trunks or Individual VLAN to each ESX host's NIC port?

I plan to connect 12 LAN patch cords from a pair of stacked Cisco
Access LAN switches to a pair of ESXi hosts.

I have 2 options:

Option 1:
a) 1 port per host to each switch's port that's assigned Management
b) 1 port per host to each switch's port that's assigned vMotion
c) the remaining 10 ports per server: 5 to each switch that has various
    data/prod/user access VLAN(s) assigned to the switch ports

Option 2:
a) all switch ports are trunked (with Management, vMotion, data/prod
    VLANs permitted to all ports on both the switches): so just need any
    6 ports on each host to each switch (as all ports on the switches are
    the same)

I think option 2 has better flexibility.

However, which of the above 2 options are better in terms of :
a) suppose 1 switch completely fails, I would like minimal downtime
    to the VMs.  I think with option 2, I've seen half of the VMs in each
    ESXi host will become unpingable for about 5 secs & then they
    become pingable again.  Any other configuration to even not to
    have this 5 secs downtime?

b) I think with Option 1, there could be an STP reconfiguration (can't
     recall the right term to use for this Spanning Tree Protocol
     that needs to recalculate itself) timing which could be 30secs?

c) which is a best practice among the 2 options & why

d) there's been a bug of ESXi 5.1 that caused crippling Unicast traffic
     flooding, so which of the 2 options can isolate/contain such issue better?
4 Solutions
sunhuxAuthor Commented:
e) is VLAN tagging essential & which of the 2 options can
    support VLAN tagging for ESXi?

f) any LACP at the switches end needed & is LACP a good
   feature to have

I suppose for Option 2, we need to set each port to "trunk ... portfast
or Option 1 needs it as well?
sunhuxAuthor Commented:
g) Suppose we're using F5 load-balancer LTM in our environment,
    which of the 2 options is more suitable?
sunhuxAuthor Commented:
Note that our pair of F5 LTM will be physically connected to the same
Access switches: one LTM to each switch
The new generation of project management tools

With monday.com’s project management tool, you can see what everyone on your team is working in a single glance. Its intuitive dashboards are customizable, so you can create systems that work for you.

BusbarSolutions ArchitectCommented:
If the switches are stacked, if one failed you will still access the VLANs over the other switch but you will lose some BW.

my suggestion create a trunk on all the ports, create switch on individual ports (via ESXi), and configure vmotion, MGMT and prod traffic on separate switches.
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
Use VLANs and reduce network cabling.

Create Trunks and run multiple VLANs over trunks, but be careful because running multiple VLANs inside trunks can be difficult to monitor performance on each individual VLAN.
I would create a trunk between each physical nic on the esx hosts to the switch stack, with each trunk on a different switch within the stack.
Then, within the esx, create virtual networks associated with each vlan which are loadbalanced between each physical adapter.

That way, it is a tidyer setup, and then you have network redundancy from the loadbalanced network adapters and the multiple switches in the stack.

I also use virtualisd and physical F5 loadbalancers in my office, and their external interface is assigned to their own vlan on the virtual. That way, I can have the f5 go between the hosts of my esx cluster without having to dedicate physical adapters on my esx hosts. Our physical doesnt have this setup, and requires to be directly attached to the firewall, so it is no where near as flexible
Mohammed KhawajaManager - Infrastructure:  Information TechnologyCommented:
My recommendation would be to create trunks for all ports.  Dedicate port 1 on each switch for management and port 2 as standby for management.  Dedicate port 2 for vmotion with port 1 for standby for vmotion.  Create a VDS with the other four to be used vm traffic.
sunhuxAuthor Commented:
Post closure query:

Will LACP help speed up the STP convergence (in the event one
of the switch is down) ?
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

Join & Write a Comment

Featured Post

Hire Technology Freelancers with Gigs

Work with freelancers specializing in everything from database administration to programming, who have proven themselves as experts in their field. Hire the best, collaborate easily, pay securely, and get projects done right.

Tackle projects and never again get stuck behind a technical roadblock.
Join Now