Learn how to a build a cloud-first strategyRegister Now

x
  • Status: Solved
  • Priority: Low
  • Security: Public
  • Views: 114
  • Last Modified:

Network configuration on VMware host

Dear Wizards, can you please suggest on this scenario?

If a physical host (with 4 physical NICs) connect to Core Switch Cisco  3925, can we use the Link Aggregation Control Protocol (LACP) to increase the bandwidth for that host? And if so, how to increase the maximum bandwidth for virtual NIC of VMs inside that host?

Many thanks in advance!
1
13L@CK_H3@RT
Asked:
13L@CK_H3@RT
  • 9
  • 9
4 Solutions
 
Ravi Kumar AtreyLead IT Infrastructure Cloud Servers and StorageCommented:
You need to first team the physical NICs on the host machine and then you can then create it as virtual NIC. Since you wants to configure LACP on cisco, then you also need to enable LACP on host machine as well in the teaming settings.

https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1004048

https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1004088
0
 
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
if this is VMware vSphere ESXi hosts LACP is only supported in Distributed Switches, not standard switches.

So a standard trunk no LACP will be sufficient just for standard switches. But do remember that physical configuration of the network switch dictates inbound traffic, outbound traffic is based on the teaming policy selected, and the two need to match for correct configuration.

All you need to do, is add a virtual machine network interface, make sure you use the VMXNET3 interface, and that has a network interface of 10GBpe - is that enough ?
0
 
13L@CK_H3@RTNetwork AdministratorAuthor Commented:
Dear Andrew, many thanks for you reply but I am not sure if I can understand you well. Could you please clarify?
- What is VMXNET3 interface?
- "virtual machine network interface" = "virtual NIC of a VM with a speed of 10 Gb/s"?

Is it required a downtime when we add new distribution switch and configure LACP on an environment of standard switch?
0
A Cyber Security RX to Protect Your Organization

Join us on December 13th for a webinar to learn how medical providers can defend against malware with a cyber security "Rx" that supports a healthy technology adoption plan for every healthcare organization.

 
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
VMXNET3 is the virtual network interface in the Guest Virtual Machine.

Firstly VMware vSphere supports two different  networking setups...

1. Standard Switches
2. Distributed Switches

LACP is just a stack/trunk protocol. You can still build four port trunks without LACP.

How many ESXi hosts do you have ?
0
 
13L@CK_H3@RTNetwork AdministratorAuthor Commented:
Hi, I have 4 ESXi hosts, each of them have 1 physical network cable connected to Core switch Cisco 3560.

currently we use standard switch setting, but seems like the read/write operations in VM are slow. Just for your information about our diagram:

HP MSA 2040 (Raid6)--------- 8 x Fibre optical ------------ 4 Hosts HP Gen9 -----------4 x 1 Gb/s ------------- Core switch Cisco 3560 ------------ Router Cisco 3925 --------- ------ Internet
0
 
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
It could be a number of reasons read/write access are SLOW!

All VMs should have VMware Tools installed and VMXNET3 interfaces installed  - that's a given.

What data and diagnosis have you performed to indicate the bottleneck is in the 4 x 1GB/s network interfaces ?
0
 
13L@CK_H3@RTNetwork AdministratorAuthor Commented:
Yes, we installed VMware tools for all 20 VMs, not sure how to check if VMXNET3 interfaces installed, could you please suggest?

Please find the attachments for the NICs, SAN and Disk info: sometimes I saw the disk speed peaked to 241 MB/s but decrease immediately. The average speed is about 60 MB/s, do you think it is too slow for a VM (RAM 80GB, 40 CPUs)?
NIC.PNG
Disk.PNG
SAN.PNG
0
 
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
Yes, we installed VMware tools for all 20 VMs, not sure how to check if VMXNET3 interfaces installed, could you please suggest?

VM Settings - check the nic ?

You've assigned a VM with 80 vCPUs ?
0
 
13L@CK_H3@RTNetwork AdministratorAuthor Commented:
No, 40 vProcessors. is it a problem?
12.PNG
0
 
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
Could be....and if you've added 40 vCPU, and if the VM needs them, it may be better that this server is physical and can use all the resources.

What was the thinking or logic of adding 40?

what is the server, and what function does it perform, to have 40 vCPU allocated to it....

vSMP (virtual SMP) can affect virtual machine performance, when adding too many vCPUs to virtual machines that cannot use the vCPUs effectively, e.g. Servers than can use vSMP correctly :- SQL Server, Exchange Server.

This is true, many VMware Administrators, think adding lots of processors, will increase performance - wrong! (and because they can, they just go silly!). Sometimes there is confusion between cores and processors. But what we are adding is additional processors in the virtual machine.

So 4 vCPU, to the VM is a 4 Way SMP (Quad Processor Server), if you have Enterprise Plus license you can add 8, (and only if you have the correct OS License will the OS recognize them all).

If applications, can take advantage e.g. Exchange, SQL, adding additional processors, can/may increase performance.

So usual rule of thumb is try 1 vCPU, then try 2 vCPU, knock back to 1 vCPU if performance is affected. and only use vSMP if the VM can take advantage.

Example, VM with 4 vCPUs allocated!

My simple laymans explaination of the "scheduler!"

As you have assigned 4 vCPUs, to this VM, the VMware scheulder, has to wait until 4 cores are free and available, to do this, it has to pause the first cores, until the 4th is available, during this timeframe, the paused cores are not available for processes, this is my simplistic view, but bottom line is adding more vCPUs to a VM, may not give you the performance benefits you think, unless the VM, it's applications are optimised for additional vCPUs.

See here
http://www.vmware.com/resources/techresources/10131

see here
http://www.gabesvirtualworld.com/how-too-many-vcpus-can-negatively-affect-your-performance/

http://www.zdnet.com/virtual-cpus-the-overprovisioning-penalty-of-vcpu-to-pcpu-ratios-4010025185/

also there is a document here about the CPU scheduler

www.vmware.com/files/pdf/perf-vsphere-cpu_scheduler.pdf

https://blogs.vmware.com/vsphere/2013/10/does-corespersocket-affect-performance.html
1
 
13L@CK_H3@RTNetwork AdministratorAuthor Commented:
Yes that one is an Exchange server 2016, serving about 900 users, the CPU utilization reduces significantly when we increase the vProcessor, and yes we have Ent plus license.

Do you think the SAN storage is the reason for slow read/write?

The high RAM in picture is due to our import data process, normally it is about 60%, CPU is about 10%
34.PNG
0
 
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
So you really think 40 virtual processors are necessary after reading my last post?
0
 
13L@CK_H3@RTNetwork AdministratorAuthor Commented:
Hi andrew, I am doing a test to measure the I/O rate when varying the vCPU of VM.

Apart from the number of vCPU, is there any other factor which can affect the I/O performance?
0
 
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
I would have a look through this EE Article I wrote

SAN storage can affect I/O
0
 
13L@CK_H3@RTNetwork AdministratorAuthor Commented:
Hi, this is the first part of a test. I will keep the RAM unchanged 20GB and varying the vCPU.

micro-benchmark is measuring the taken time to run command 1000 times. Repeat 100 times and calculate the average value. Do you think this test is suitable?

This is the command on PowerShell:
for ($n=1;$n -le 100;$n++) {Measure-Command {for($i=1; $i -le 1000; $i++){echo Hello world}} | find "TotalMilliseconds :"}

Open in new window

test.PNG
0
 
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
by using a VM with 40 vCPUs will affect performance on all the other VMs.

What is the host ?

and if you did need a VM which needed so much resource why would your virtualise it.

I would look at performance overall on the host.
0
 
13L@CK_H3@RTNetwork AdministratorAuthor Commented:
please see the attachment for the host's information.

Because we have to deploy Exchange system for 6 subsidiaries and some other servers so we choose VMware to save money. The VM which have 40 vCPU is standing alone in one host. The VM which I am testing on is in other host (in the picture)
esxi.PNG
0
 
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
not sure how you are saving money, if a VM has all the resources associated with it, you might as well remove the hypervisor, and have the OS on Bare Metal accessing all the resources as the hypervisor will slow the VM down!

I think you seriously need to look at the VM requirements for this Exchange Server, it's not benefiting from Virtualisation by assigning all the vCPUs to it.

It's possibly CPU oversubscribed and running slower than it should.

see here

http://msexchangeguru.com/2016/06/07/exch-proc-calc/
0
 
13L@CK_H3@RTNetwork AdministratorAuthor Commented:
Many thanks for your answers.
0

Featured Post

New Tabletop Appliances Blow Competitors Away!

WatchGuard’s new T15, T35 and T55 tabletop UTMs provide the highest-performing security inspection in their class, allowing users at small offices, home offices and distributed enterprises to experience blazing-fast Internet speeds without sacrificing enterprise-grade security.

  • 9
  • 9
Tackle projects and never again get stuck behind a technical roadblock.
Join Now