Hi Kirk, Good day to you, and thank you for your e-mail.
Yes, I have confirmed that the compute offering is causing this. The network rate on the compute offering for all the VMs were set to 2 Mbit/sec. For some reason, the network rate was not enforced when I was still using CloudStack 4.1.1 and earlier, but now being enforced on CloudStack 4.2.0. Thank you for your detailed information on compute offering. I have tested by creating a new compute offering with higher network rate (100 Mbit/sec), shutdown the instance, change service offering to the new compute offering, and start back the instance. I am able to get much higher bandwidth during my speedtest now. Thank you. On Fri, Oct 4, 2013 at 3:12 PM, Kirk Kosinski <kirkkosin...@gmail.com>wrote: > Hi, your observations on Compute Offerings are by design. They can only > be created and managed by admin accounts, and only the name and > description can be edited. They can be public or domain-specific, but a > root admin account should be able to view all of them. > > Instead of editing existing offerings to set the desired network > throttling, new ones need to be created (using an admin account) and the > VMs updated. VMs will need to be stopped, and it can be done by a root > admin or the user that owns the VMs. The change can be made via UI or > API (changeServiceForVirtualMachine command). > > But are you sure the Compute Offering is causing this? Is the Network > Rate on the Compute Offering for the affected VMs set to 2 Mbit/sec? > What hypervisor are the affected VMs running on, and what type of > network (isolated or shared)? The actual network throttling seen by VMs > is complicated and depends on a combination of factors including: > > 1) Network type (isolate or shared) > 2) Hypervisor of the guest > 3) Hypervisor of the virtual router (for isolated networks) > 4) NIC driver of virtual router (for vSphere) > 5) Network Offering for the network > 6) Compute Offering for the instance > 7) System Service Offering for the virtual router (isolated networks) > 8) Direction of traffic (ingress or egress) > 9) The vm.network.throttling.rate global parameter > 10) The network.throttling.rate zone parameter > 11) Actual physical capabilities of the network and hosts > > Best regards, > Kirk > > On 10/03/2013 08:26 PM, Indra Pramana wrote: > > Dear all, > > > > After CloudStack upgrade to 4.2.0, the network speed to our VM instances > is > > very slow and seems to be limited to 2 Mbps for both ingress and egress > > traffic for each instance. It seems that there's some network rate > limiting > > which didn't take effect on CloudStack version 4.1.1 or below, but now is > > taking effect on CloudStack 4.2.0. > > > > I read this article on how CloudStack manages the network traffic for the > > VMs: > > > > > http://cloudstack.apache.org/docs/en-US/Apache_CloudStack/4.1.1/html/Admin_Guide/network-rate.html > > > > For guest VMs, the network rate-limiting is set on the compute offering > > associate with each VM. Please CMIIW. > > > > All our guest VMs are created under different accounts and have their own > > compute offering, which is auto-created under that specific account when > > the VM instance is created. > > > > The issue is, from admin account, I can't seem to modify the compute > > offering for the guest VMs. If I go to GUI > Service Offerings > Compute > > Offering, I can only see the compute offering for admin account only. How > > can I see and modify the compute offering under different accounts? > > > > Furthermore, it seems that I cannot modify the network rate field of an > > existing compute offering. How do I modify it so that I can set higher > > network rate limit or unlimited bandwidth for an existing guest VM? > > > > Looking forward to your reply, thank you. > > > > Cheers. > > >
