Recommended to turn off LSO (Large Segment Offload) in Solaris x86 VMs in vSphere 5.0 ?

Is it recommended to turn off the LSO (Large Segment Offload) in solaris x86 VMs that run in vSphere 5.0 Upd 3
cluster?  We are concerned that this may affect the max size of files being copied across VMs & performance

To disable LSO in Solaris,
ndd -set /dev/ip ip_lso_outbound 0

Reference:
http://blog.cyberexplorer.me/2013/03/improving-vm-to-vm-network-throughput.html
sunhuxAsked:
Who is Participating?

[Product update] Infrastructure Analysis Tool is now available with Business Accounts.Learn More

x
I wear a lot of hats...

"The solutions and answers provided on Experts Exchange have been extremely helpful to me over the last few years. I wear a lot of hats - Developer, Database Administrator, Help Desk, etc., so I know a lot of things but not a lot about one thing. Experts Exchange gives me answers from people who do know a lot about one thing, in a easy to use platform." -Todd S.

sunhuxAuthor Commented:
Do share any VMWare or Oracle KB article if any
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
Same question, are you using the VMXNET3 interface.
sunhuxAuthor Commented:
Yes, we are using VMXNET3 : is it only with Win2008 R2 or with other OS (like Solaris x86,
Win2012 & RHEL 5.x/6.x ? )
Active Protection takes the fight to cryptojacking

While there were several headline-grabbing ransomware attacks during in 2017, another big threat started appearing at the same time that didn’t get the same coverage – illicit cryptomining.

gheistCommented:
Best practice is to not try to fix what is not broken.
giltjrCommented:
How do you think turning off LSO will affect the size of a file being copied?  All this does is move where the packets are segment down to the MTU.
sunhuxAuthor Commented:
My colleague found an article on VMXNET3 vNICs that would stop
large files download unless "auto-tuning" is turned off but this is
for Windows 2008 R2 VMs that he had tested after disabling
auto-tuning, he could successfully copy over large files;  so he's
asking for the equivalent of 'auto-tuning' for Solaris x86 (which
he reckons is LSO)
Autotune-network-performance-with-vmxnet
gheistCommented:
very non-official source.
Yes, sometiems vmware's drivers are buggy, then first thing to try is to change to e1000 (and open support case with vmware)
giltjrCommented:
Agree with Gheist, if there is a some sort of bug it could cause lots of problems.  I would say that its not just "large file downloads" but possibly any time a large amount of data is being transferred in a single stream.  Such as streaming video or audio.
sunhuxAuthor Commented:
Yes, any large amt of data in a single stream will be dropped & we faced this not with just
a couple of VMs but dozens of VMs.

Saw some KB articles that E1000 has issues too, in fact more documented E1000 issues
than with vmxnet3.

So we have to disable 'auto-tuning' or LSO as a workaround?
gheistCommented:
Is it dropped or not? What will happen you cannot know now.
e.g. ESX/i 5U3 fixes bug where packets are lost wildly on same host.
sunhuxAuthor Commented:
I don't know exactly what happened but large files transfer tend to 'freeze'.

How about article below?  Is it a less updated article?

http://blog.cyberexplorer.me/2013/03/improving-vm-to-vm-network-throughput.html
gheistCommented:
Large send offload is something about sending 64k to the netcard which splits it into 9k or 1.5k frames on the wire. It has no influence whatsoever on file transfer sizes.
Please update to 5.0U3 and check again IF there is any issue.
giltjrCommented:
In a virtual world I don't think LSO buys you anything.  

What LSO does is allows the IP stack to send a packet larger than the network's MTU to the NIC.  The NIC will then break the packet down to the network's MTU size (typically 1500 or 1492).  This offloads some of the processing from the CPU to the NIC.

The issue here is that this is a virtual machine, so the NIC is not a physical device in this situation, but software.  So there is no real processing offload.  It just move from being handled by the VM Guest  to being handled by the hypervisor.   So I don't think there is any real CPU saving.  The guest might see some CPU usage reduction, but that would show up as CPU usage increase in the hypervisor.

Experts Exchange Solution brought to you by

Your issues matter to us.

Facing a tech roadblock? Get the help and guidance you need from experienced professionals who care. Ask your question anytime, anywhere, with no hassle.

Start your 7-day free trial
sunhuxAuthor Commented:
We are already on Ver 5.0 Upd 3
gheistCommented:
Can you catch vmware red-handed eating up your packets? Yes - support case No - do not fix working system
btw netalyzr (java CLI) can tell you if fragments are eaten up vs normal physical machine.
It's more than this solution.Get answers and train to solve all your tech problems - anytime, anywhere.Try it for free Edge Out The Competitionfor your dream job with proven skills and certifications.Get started today Stand Outas the employee with proven skills.Start learning today for free Move Your Career Forwardwith certification training in the latest technologies.Start your trial today
VMware

From novice to tech pro — start learning today.