Hello guys,
I'm testing an preproduction environment with cs4.0.0 onto centos 5.3.
I'm testing a vpc environment.
I'm asking why does the tier networks are limited to 200mb/s (network
offering DefaultIsolatedNetworkOfferingForVpcNetworksNoLB) ?
So in xencenter, if i look at the network tab of m
hi,
the default value of network.throttling.rate and
vm.network.throttling.rate is 200mbps, you can change it in global
setting. Restart management server and network.
best,
wei
2013/4/17, benoit lair :
> Hello guys,
>
>
> I'm testing an preproduction environment with cs4.0.0 onto centos 5.3.
>
>
You can change it on the network offering level (networkRate parameter).
Create a new network offering with the desired network rate, and upgrade
your existing network to the new offering. The advantage of this approach
over the global config parameter change is - the global config change will
affe
yes, you can limit network rate of certain type of networks according to
Alena's approach.
Maybe we also need to limit the network rate of "VMs" in certain networks?
2013/4/17 Alena Prokharchyk
> You can change it on the network offering level (networkRate parameter).
> Create a new network off
I've located the detailed article on how to setup the network throttling
in cloudStack:
http://support.citrix.com/article/CTX132019
The approaches base on the Network type (Guest,Public,Control) and vmType
(VR/UserVm/SSVM)
We might need to add it to Apache Wiki.
-Alena.
On 4/17/13 11:09 AM, "
Alena, Thank you so much for your reply.
It is very clear and helpful for me.
I will work on vm disk throttling in the next days.
2013/4/17 Alena Prokharchyk
> I've located the detailed article on how to setup the network throttling
> in cloudStack:
>
> http://support.citrix.com/article/CTX1320
Hello,
Thanks for your response.
However, do you know in production cases, what are the maximum bandwidth i
can manage with a vpc vr ? Are there limitations or any recommendations
(sizing limit in order to avoid performance loss) concerning provisionning
of the vpc ?
Thanks for your advices.
Really depends on
- your CPU configuration
- your hypervisor
- your physical network speed
- noisy neighbors around the VR / around the physical network
- how many ipsec vpns
- how many LB
- load on the LB
- number of connections
If you just test the VR in isolation with a 512GB RAM/1core
Hello Chiradeep,
So with a VR running on a x5550 hexacore with not enough noisy neighboors
with a computing offering with 4 core, 4 gb ram, i can hope to run 20 vlans
with a total troughput of 2 gb max with 10 ip load balanced (each onto 3
vms) for a total amount of 3k connections (max, avg at 500
Increasing cores won't get you much since the current VR (4.1) linux
2.6.32 where the networking tends to be single-core only. Otherwise YMMV !
Increasing memory also does not give you much -- the bottleneck is the
memory used to store conn track connections -- 512mB will get you about 1
million co
Thanks for your response Chiradeep.
2013/4/19 Chiradeep Vittal
> Increasing cores won't get you much since the current VR (4.1) linux
> 2.6.32 where the networking tends to be single-core only. Otherwise YMMV !
> Increasing memory also does not give you much -- the bottleneck is the
> memory us
11 matches
Mail list logo
