Hi all, I have install xen 3.4.3 based on centos5.4. I changed a line in /etc/xen/xend-config.sxp like below: (network-script network-bridge) ----> (network-script my-network-bridge) and #cat /etc/xen/scripts/my-network-script prints as below: #!/bin/sh /etc/xen/scripts/network-bridge "$@" netdev=eth0 bridge=eth0 /etc/xen/scripts/network-bridge "$@" netdev=eth1 bridge=eth1 Now I installed two vms, and set one vm using eth0, then set the other using eth1. So, I can connect internet in both vms. But now my problem is that the throughput of two network card is the same as only one network card. Another words, If I use the only one network card such as eth0 for the two vms, the throughput of eth0 is about 250m. But when I set one vm use eth0 and the other use eth1, the total throughput of eth0 and eth1 is about 250m too. That why? Any bottleneck? Thanks a lot. Best Wishes! Qiang Ma _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Pasi Kärkkäinen
2010-Sep-06 06:08 UTC
Re: [Xen-users] The bottleneck of using multinetwork card?
On Mon, Sep 06, 2010 at 01:58:10PM +0800, ma qiang wrote:> Hi all, > I have install xen 3.4.3 based on centos5.4. > I changed a line in /etc/xen/xend-config.sxp like below: > (network-script network-bridge) ----> (network-script my-network-bridge) > > and #cat /etc/xen/scripts/my-network-script prints as below: > #!/bin/sh > /etc/xen/scripts/network-bridge "$@" netdev=eth0 bridge=eth0 > /etc/xen/scripts/network-bridge "$@" netdev=eth1 bridge=eth1 > > Now I installed two vms, and set one vm using eth0, then set the other > using eth1. > So, I can connect internet in both vms. > > But now my problem is that the throughput of two network card is the > same as only one network card. > Another words, If I use the only one network card such as eth0 for the > two vms, the throughput of eth0 is about 250m. > But when I set one vm use eth0 and the other use eth1, the total > throughput of eth0 and eth1 is about 250m too. > > That why? Any bottleneck? > Thanks a lot. >Do you mean 250 Mbit/sec ? You should get much more.. how are you measuring the performance? What benchmark are you using? Are both the eth0 and eth1 connected to the same network/vlan? When you run the network benchmark try running "xm top" to monitor the overall cpu usage.. also run "top" in dom0 to see if there''s a big cpu usage in dom0. -- Pasi _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
> Do you mean 250 Mbit/sec ?*Yes.* * *> You should get much more.. how are you measuring the performance?> What benchmark are you using?*I use an tool from ixia to test the network in the vm.*> Are both the eth0 and eth1 connected to the same network/vlan?*Yes. I run my test from another computer to connect to vm0 and vm1, and they all in the same switch. * *To be noted my cards are both 1000m network card.*> When you run the network benchmark try running "xm top" > to monitor the overall cpu usage.. also run "top" in dom0 > to see if there''s a big cpu usage in dom0.*No, the cpu is less than 20%* On Mon, Sep 6, 2010 at 2:08 PM, Pasi Kärkkäinen <pasik@iki.fi> wrote:> On Mon, Sep 06, 2010 at 01:58:10PM +0800, ma qiang wrote: >> Hi all, >> I have install xen 3.4.3 based on centos5.4. >> I changed a line in /etc/xen/xend-config.sxp like below: >> (network-script network-bridge) ----> (network-scriptmy-network-bridge)>> >> and #cat /etc/xen/scripts/my-network-script prints as below: >> #!/bin/sh >> /etc/xen/scripts/network-bridge "$@" netdev=eth0 bridge=eth0 >> /etc/xen/scripts/network-bridge "$@" netdev=eth1 bridge=eth1 >> >> Now I installed two vms, and set one vm using eth0, then set the other >> using eth1. >> So, I can connect internet in both vms. >> >> But now my problem is that the throughput of two network card is the >> same as only one network card. >> Another words, If I use the only one network card such as eth0 for the >> two vms, the throughput of eth0 is about 250m. >> But when I set one vm use eth0 and the other use eth1, the total >> throughput of eth0 and eth1 is about 250m too. >> >> That why? Any bottleneck? >> Thanks a lot. >> > > Do you mean 250 Mbit/sec ? > > You should get much more.. how are you measuring the performance? > What benchmark are you using? > > Are both the eth0 and eth1 connected to the same network/vlan? > > When you run the network benchmark try running "xm top" > to monitor the overall cpu usage.. also run "top" in dom0 > to see if there''s a big cpu usage in dom0. > > -- Pasi > >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Pasi Kärkkäinen
2010-Sep-06 06:50 UTC
Re: [Xen-users] The bottleneck of using multinetwork card?
On Mon, Sep 06, 2010 at 02:44:59PM +0800, ma qiang wrote:> > Do you mean 250 Mbit/sec ? > Yes. > > You should get much more.. how are you measuring the performance? > > What benchmark are you using? > I use an tool from ixia to test the network in the vm.Try using some standard network performance testing tools such as "iperf" or just regular ftp file transfers.> > Are both the eth0 and eth1 connected to the same network/vlan? > Yes. I run my test from another computer to connect to vm0 and vm1, and > they all in the same switch. > To be noted my cards are both 1000m network card. >How is the ''test server'' connected?> > When you run the network benchmark try running "xm top" > > to monitor the overall cpu usage.. also run "top" in dom0 > > to see if there''s a big cpu usage in dom0. > No, the cpu is less than 20%20% on dom0? How about the domUs? Do they have big cpu usage? What kind of guests do you have? pv? hvm? linux? windows? -- Pasi> On Mon, Sep 6, 2010 at 2:08 PM, Pasi Kärkkäinen <[1]pasik@iki.fi> wrote: > > On Mon, Sep 06, 2010 at 01:58:10PM +0800, ma qiang wrote: > >> Hi all, > >> I have install xen 3.4.3 based on centos5.4. > >> I changed a line in /etc/xen/xend-config.sxp like below: > >> (network-script network-bridge) ----> (network-script > my-network-bridge) > >> > >> and #cat /etc/xen/scripts/my-network-script prints as below: > >> #!/bin/sh > >> /etc/xen/scripts/network-bridge "$@" netdev=eth0 bridge=eth0 > >> /etc/xen/scripts/network-bridge "$@" netdev=eth1 bridge=eth1 > >> > >> Now I installed two vms, and set one vm using eth0, then set the other > >> using eth1. > >> So, I can connect internet in both vms. > >> > >> But now my problem is that the throughput of two network card is the > >> same as only one network card. > >> Another words, If I use the only one network card such as eth0 for the > >> two vms, the throughput of eth0 is about 250m. > >> But when I set one vm use eth0 and the other use eth1, the total > >> throughput of eth0 and eth1 is about 250m too. > >> > >> That why? Any bottleneck? > >> Thanks a lot. > >> > > > > Do you mean 250 Mbit/sec ? > > > > You should get much more.. how are you measuring the performance? > > What benchmark are you using? > > > > Are both the eth0 and eth1 connected to the same network/vlan? > > > > When you run the network benchmark try running "xm top" > > to monitor the overall cpu usage.. also run "top" in dom0 > > to see if there''s a big cpu usage in dom0. > > > > -- Pasi > > > > > > References > > Visible links > 1. mailto:pasik@iki.fi_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
I did the same test one the same computer in dom0 using only one network card, the results show 1000M/sec. On Mon, Sep 6, 2010 at 2:44 PM, ma qiang <maqiang1984@gmail.com> wrote:> > Do you mean 250 Mbit/sec ? > *Yes.* > * > *> You should get much more.. how are you measuring the performance? > > What benchmark are you using? > *I use an tool from ixia to test the network in the vm.* > > > Are both the eth0 and eth1 connected to the same network/vlan? > *Yes. I run my test from another computer to connect to vm0 and vm1, and > they all in the same switch. * > *To be noted my cards are both 1000m network card.* > > > When you run the network benchmark try running "xm top" > > to monitor the overall cpu usage.. also run "top" in dom0 > > to see if there''s a big cpu usage in dom0. > *No, the cpu is less than 20%* > > On Mon, Sep 6, 2010 at 2:08 PM, Pasi Kärkkäinen <pasik@iki.fi> wrote: > > On Mon, Sep 06, 2010 at 01:58:10PM +0800, ma qiang wrote: > >> Hi all, > >> I have install xen 3.4.3 based on centos5.4. > >> I changed a line in /etc/xen/xend-config.sxp like below: > >> (network-script network-bridge) ----> (network-script > my-network-bridge) > >> > >> and #cat /etc/xen/scripts/my-network-script prints as below: > >> #!/bin/sh > >> /etc/xen/scripts/network-bridge "$@" netdev=eth0 bridge=eth0 > >> /etc/xen/scripts/network-bridge "$@" netdev=eth1 bridge=eth1 > >> > >> Now I installed two vms, and set one vm using eth0, then set the other > >> using eth1. > >> So, I can connect internet in both vms. > >> > >> But now my problem is that the throughput of two network card is the > >> same as only one network card. > >> Another words, If I use the only one network card such as eth0 for the > >> two vms, the throughput of eth0 is about 250m. > >> But when I set one vm use eth0 and the other use eth1, the total > >> throughput of eth0 and eth1 is about 250m too. > >> > >> That why? Any bottleneck? > >> Thanks a lot. > >> > > > > Do you mean 250 Mbit/sec ? > > > > You should get much more.. how are you measuring the performance? > > What benchmark are you using? > > > > Are both the eth0 and eth1 connected to the same network/vlan? > > > > When you run the network benchmark try running "xm top" > > to monitor the overall cpu usage.. also run "top" in dom0 > > to see if there''s a big cpu usage in dom0. > > > > -- Pasi > > > > > >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Pasi Kärkkäinen
2010-Sep-06 06:56 UTC
Re: [Xen-users] The bottleneck of using multinetwork card?
On Mon, Sep 06, 2010 at 02:52:16PM +0800, ma qiang wrote:> I did the same test one the same computer in dom0 using only one network > card, the results show 1000M/sec. >So what kind of guests are you using the benchmark from? -- Pasi> On Mon, Sep 6, 2010 at 2:44 PM, ma qiang <[1]maqiang1984@gmail.com> wrote: > > > Do you mean 250 Mbit/sec ? > Yes. > > You should get much more.. how are you measuring the performance? > > What benchmark are you using? > I use an tool from ixia to test the network in the vm. > > Are both the eth0 and eth1 connected to the same network/vlan? > Yes. I run my test from another computer to connect to vm0 and vm1, > and they all in the same switch. > To be noted my cards are both 1000m network card. > > When you run the network benchmark try running "xm top" > > to monitor the overall cpu usage.. also run "top" in dom0 > > to see if there''s a big cpu usage in dom0. > No, the cpu is less than 20% > On Mon, Sep 6, 2010 at 2:08 PM, Pasi Kärkkäinen <[2]pasik@iki.fi> wrote: > > On Mon, Sep 06, 2010 at 01:58:10PM +0800, ma qiang wrote: > >> Hi all, > >> I have install xen 3.4.3 based on centos5.4. > >> I changed a line in /etc/xen/xend-config.sxp like below: > >> (network-script network-bridge) ----> (network-script > my-network-bridge) > >> > >> and #cat /etc/xen/scripts/my-network-script prints as below: > >> #!/bin/sh > >> /etc/xen/scripts/network-bridge "$@" netdev=eth0 bridge=eth0 > >> /etc/xen/scripts/network-bridge "$@" netdev=eth1 bridge=eth1 > >> > >> Now I installed two vms, and set one vm using eth0, then set the > other > >> using eth1. > >> So, I can connect internet in both vms. > >> > >> But now my problem is that the throughput of two network card is the > >> same as only one network card. > >> Another words, If I use the only one network card such as eth0 for > the > >> two vms, the throughput of eth0 is about 250m. > >> But when I set one vm use eth0 and the other use eth1, the total > >> throughput of eth0 and eth1 is about 250m too. > >> > >> That why? Any bottleneck? > >> Thanks a lot. > >> > > > > Do you mean 250 Mbit/sec ? > > > > You should get much more.. how are you measuring the performance? > > What benchmark are you using? > > > > Are both the eth0 and eth1 connected to the same network/vlan? > > > > When you run the network benchmark try running "xm top" > > to monitor the overall cpu usage.. also run "top" in dom0 > > to see if there''s a big cpu usage in dom0. > > > > -- Pasi > > > > > > References > > Visible links > 1. mailto:maqiang1984@gmail.com > 2. mailto:pasik@iki.fi_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
20% is the vm''s utilization of cpu, not dom0''s. My domU is hvm and centos linux. "test server" is OK, and I have tested is on the same computer using only one network card. the results show 1000M. On Mon, Sep 6, 2010 at 2:50 PM, Pasi Kärkkäinen <pasik@iki.fi> wrote:> On Mon, Sep 06, 2010 at 02:44:59PM +0800, ma qiang wrote: > > > Do you mean 250 Mbit/sec ? > > Yes. > > > You should get much more.. how are you measuring the performance? > > > What benchmark are you using? > > I use an tool from ixia to test the network in the vm. > > Try using some standard network performance testing tools > such as "iperf" or just regular ftp file transfers. > > > > Are both the eth0 and eth1 connected to the same network/vlan? > > Yes. I run my test from another computer to connect to vm0 and vm1, > and > > they all in the same switch. > > To be noted my cards are both 1000m network card. > > > > How is the ''test server'' connected? > > > > > When you run the network benchmark try running "xm top" > > > to monitor the overall cpu usage.. also run "top" in dom0 > > > to see if there''s a big cpu usage in dom0. > > No, the cpu is less than 20% > > 20% on dom0? > How about the domUs? Do they have big cpu usage? > > What kind of guests do you have? pv? hvm? linux? windows? > > -- Pasi > > > On Mon, Sep 6, 2010 at 2:08 PM, Pasi Kärkkäinen <[1]pasik@iki.fi> > wrote: > > > On Mon, Sep 06, 2010 at 01:58:10PM +0800, ma qiang wrote: > > >> Hi all, > > >> I have install xen 3.4.3 based on centos5.4. > > >> I changed a line in /etc/xen/xend-config.sxp like below: > > >> (network-script network-bridge) ----> (network-script > > my-network-bridge) > > >> > > >> and #cat /etc/xen/scripts/my-network-script prints as below: > > >> #!/bin/sh > > >> /etc/xen/scripts/network-bridge "$@" netdev=eth0 bridge=eth0 > > >> /etc/xen/scripts/network-bridge "$@" netdev=eth1 bridge=eth1 > > >> > > >> Now I installed two vms, and set one vm using eth0, then set the > other > > >> using eth1. > > >> So, I can connect internet in both vms. > > >> > > >> But now my problem is that the throughput of two network card is > the > > >> same as only one network card. > > >> Another words, If I use the only one network card such as eth0 for > the > > >> two vms, the throughput of eth0 is about 250m. > > >> But when I set one vm use eth0 and the other use eth1, the total > > >> throughput of eth0 and eth1 is about 250m too. > > >> > > >> That why? Any bottleneck? > > >> Thanks a lot. > > >> > > > > > > Do you mean 250 Mbit/sec ? > > > > > > You should get much more.. how are you measuring the performance? > > > What benchmark are you using? > > > > > > Are both the eth0 and eth1 connected to the same network/vlan? > > > > > > When you run the network benchmark try running "xm top" > > > to monitor the overall cpu usage.. also run "top" in dom0 > > > to see if there''s a big cpu usage in dom0. > > > > > > -- Pasi > > > > > > > > > > References > > > > Visible links > > 1. mailto:pasik@iki.fi >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Sorry, I lost two words. I reboot in bare linux not dom0, and have tested it on the same computer using only one network card. the results show 1000M. On Mon, Sep 6, 2010 at 2:57 PM, ma qiang <maqiang1984@gmail.com> wrote:> 20% is the vm''s utilization of cpu, not dom0''s. > My domU is hvm and centos linux. > "test server" is OK, and I have tested is on the same computer using only > one network card. the results show 1000M. > > > On Mon, Sep 6, 2010 at 2:50 PM, Pasi Kärkkäinen <pasik@iki.fi> wrote: > >> On Mon, Sep 06, 2010 at 02:44:59PM +0800, ma qiang wrote: >> > > Do you mean 250 Mbit/sec ? >> > Yes. >> > > You should get much more.. how are you measuring the performance? >> > > What benchmark are you using? >> > I use an tool from ixia to test the network in the vm. >> >> Try using some standard network performance testing tools >> such as "iperf" or just regular ftp file transfers. >> >> > > Are both the eth0 and eth1 connected to the same network/vlan? >> > Yes. I run my test from another computer to connect to vm0 and vm1, >> and >> > they all in the same switch. >> > To be noted my cards are both 1000m network card. >> > >> >> How is the ''test server'' connected? >> >> >> > > When you run the network benchmark try running "xm top" >> > > to monitor the overall cpu usage.. also run "top" in dom0 >> > > to see if there''s a big cpu usage in dom0. >> > No, the cpu is less than 20% >> >> 20% on dom0? >> How about the domUs? Do they have big cpu usage? >> >> What kind of guests do you have? pv? hvm? linux? windows? >> >> -- Pasi >> >> > On Mon, Sep 6, 2010 at 2:08 PM, Pasi Kärkkäinen <[1]pasik@iki.fi> >> wrote: >> > > On Mon, Sep 06, 2010 at 01:58:10PM +0800, ma qiang wrote: >> > >> Hi all, >> > >> I have install xen 3.4.3 based on centos5.4. >> > >> I changed a line in /etc/xen/xend-config.sxp like below: >> > >> (network-script network-bridge) ----> (network-script >> > my-network-bridge) >> > >> >> > >> and #cat /etc/xen/scripts/my-network-script prints as below: >> > >> #!/bin/sh >> > >> /etc/xen/scripts/network-bridge "$@" netdev=eth0 bridge=eth0 >> > >> /etc/xen/scripts/network-bridge "$@" netdev=eth1 bridge=eth1 >> > >> >> > >> Now I installed two vms, and set one vm using eth0, then set the >> other >> > >> using eth1. >> > >> So, I can connect internet in both vms. >> > >> >> > >> But now my problem is that the throughput of two network card is >> the >> > >> same as only one network card. >> > >> Another words, If I use the only one network card such as eth0 for >> the >> > >> two vms, the throughput of eth0 is about 250m. >> > >> But when I set one vm use eth0 and the other use eth1, the total >> > >> throughput of eth0 and eth1 is about 250m too. >> > >> >> > >> That why? Any bottleneck? >> > >> Thanks a lot. >> > >> >> > > >> > > Do you mean 250 Mbit/sec ? >> > > >> > > You should get much more.. how are you measuring the performance? >> > > What benchmark are you using? >> > > >> > > Are both the eth0 and eth1 connected to the same network/vlan? >> > > >> > > When you run the network benchmark try running "xm top" >> > > to monitor the overall cpu usage.. also run "top" in dom0 >> > > to see if there''s a big cpu usage in dom0. >> > > >> > > -- Pasi >> > > >> > > >> > >> > References >> > >> > Visible links >> > 1. mailto:pasik@iki.fi >> > >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Pasi Kärkkäinen
2010-Sep-06 07:02 UTC
Re: [Xen-users] The bottleneck of using multinetwork card?
On Mon, Sep 06, 2010 at 02:58:32PM +0800, ma qiang wrote:> Sorry, I lost two words. > I reboot in bare linux not dom0, and have tested it on the same computer > using only one network card. the results show 1000M. >So what do you get from xen dom0? Also if your guest is HVM, have you installed PV-on-HVM drivers? without those drivers the performance will be bad from the guest. -- Pasi> On Mon, Sep 6, 2010 at 2:57 PM, ma qiang <[1]maqiang1984@gmail.com> wrote: > > 20% is the vm''s utilization of cpu, not dom0''s. > My domU is hvm and centos linux. > "test server" is OK, and I have tested is on the same computer using > only one network card. the results show 1000M. > > On Mon, Sep 6, 2010 at 2:50 PM, Pasi Kärkkäinen <[2]pasik@iki.fi> wrote: > > On Mon, Sep 06, 2010 at 02:44:59PM +0800, ma qiang wrote: > > > Do you mean 250 Mbit/sec ? > > Yes. > > > You should get much more.. how are you measuring the > performance? > > > What benchmark are you using? > > I use an tool from ixia to test the network in the vm. > > Try using some standard network performance testing tools > such as "iperf" or just regular ftp file transfers. > > > Are both the eth0 and eth1 connected to the same network/vlan? > > Yes. I run my test from another computer to connect to vm0 and > vm1, and > > they all in the same switch. > > To be noted my cards are both 1000m network card. > > > > How is the ''test server'' connected? > > > > When you run the network benchmark try running "xm top" > > > to monitor the overall cpu usage.. also run "top" in dom0 > > > to see if there''s a big cpu usage in dom0. > > No, the cpu is less than 20% > > 20% on dom0? > How about the domUs? Do they have big cpu usage? > > What kind of guests do you have? pv? hvm? linux? windows? > > -- Pasi > > On Mon, Sep 6, 2010 at 2:08 PM, Pasi Kärkkäinen > <[1][3]pasik@iki.fi> wrote: > > > On Mon, Sep 06, 2010 at 01:58:10PM +0800, ma qiang wrote: > > >> Hi all, > > >> I have install xen 3.4.3 based on centos5.4. > > >> I changed a line in /etc/xen/xend-config.sxp like below: > > >> (network-script network-bridge) ----> (network-script > > my-network-bridge) > > >> > > >> and #cat /etc/xen/scripts/my-network-script prints as below: > > >> #!/bin/sh > > >> /etc/xen/scripts/network-bridge "$@" netdev=eth0 bridge=eth0 > > >> /etc/xen/scripts/network-bridge "$@" netdev=eth1 bridge=eth1 > > >> > > >> Now I installed two vms, and set one vm using eth0, then set > the other > > >> using eth1. > > >> So, I can connect internet in both vms. > > >> > > >> But now my problem is that the throughput of two network card > is the > > >> same as only one network card. > > >> Another words, If I use the only one network card such as eth0 > for the > > >> two vms, the throughput of eth0 is about 250m. > > >> But when I set one vm use eth0 and the other use eth1, the > total > > >> throughput of eth0 and eth1 is about 250m too. > > >> > > >> That why? Any bottleneck? > > >> Thanks a lot. > > >> > > > > > > Do you mean 250 Mbit/sec ? > > > > > > You should get much more.. how are you measuring the > performance? > > > What benchmark are you using? > > > > > > Are both the eth0 and eth1 connected to the same network/vlan? > > > > > > When you run the network benchmark try running "xm top" > > > to monitor the overall cpu usage.. also run "top" in dom0 > > > to see if there''s a big cpu usage in dom0. > > > > > > -- Pasi > > > > > > > > > > References > > > > Visible links > > 1. mailto:[4]pasik@iki.fi > > References > > Visible links > 1. mailto:maqiang1984@gmail.com > 2. mailto:pasik@iki.fi > 3. mailto:pasik@iki.fi > 4. mailto:pasik@iki.fi_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
>On Mon, Sep 06, 2010 at 02:52:16PM +0800, ma qiang wrote: > I did the same test one the same computer in dom0 using only onenetwork> card, the results show 1000M/sec. > >So what kind of guests are you using the benchmark from?*That is IXIA side based on centos 5.4 also.* On Mon, Sep 6, 2010 at 3:02 PM, Pasi Kärkkäinen <pasik@iki.fi> wrote:> On Mon, Sep 06, 2010 at 02:58:32PM +0800, ma qiang wrote: > > Sorry, I lost two words. > > I reboot in bare linux not dom0, and have tested it on the same > computer > > using only one network card. the results show 1000M. > > > > So what do you get from xen dom0? > *In dom0 and bare linux, the results show both 1000M. But in domU, the > results are bad.* >> Also if your guest is HVM, have you installed PV-on-HVM drivers? > without those drivers the performance will be bad from the guest. > *PV-on-HVM drivers ? Sorry, I haven''t installed that. And I''m going to **search > that.*> -- Pasi > > > On Mon, Sep 6, 2010 at 2:57 PM, ma qiang <[1]maqiang1984@gmail.com> > wrote: > > > > 20% is the vm''s utilization of cpu, not dom0''s. > > My domU is hvm and centos linux. > > "test server" is OK, and I have tested is on the same computer using > > only one network card. the results show 1000M. > > > > On Mon, Sep 6, 2010 at 2:50 PM, Pasi Kärkkäinen <[2]pasik@iki.fi> > wrote: > > > > On Mon, Sep 06, 2010 at 02:44:59PM +0800, ma qiang wrote: > > > > Do you mean 250 Mbit/sec ? > > > Yes. > > > > You should get much more.. how are you measuring the > > performance? > > > > What benchmark are you using? > > > I use an tool from ixia to test the network in the vm. > > > > Try using some standard network performance testing tools > > such as "iperf" or just regular ftp file transfers. > > > > Are both the eth0 and eth1 connected to the same > network/vlan? > > > Yes. I run my test from another computer to connect to vm0 > and > > vm1, and > > > they all in the same switch. > > > To be noted my cards are both 1000m network card. > > > > > > > How is the ''test server'' connected? > > > > > > When you run the network benchmark try running "xm top" > > > > to monitor the overall cpu usage.. also run "top" in dom0 > > > > to see if there''s a big cpu usage in dom0. > > > No, the cpu is less than 20% > > > > 20% on dom0? > > How about the domUs? Do they have big cpu usage? > > > > What kind of guests do you have? pv? hvm? linux? windows? > > > > -- Pasi > > > On Mon, Sep 6, 2010 at 2:08 PM, Pasi Kärkkäinen > > <[1][3]pasik@iki.fi> wrote: > > > > On Mon, Sep 06, 2010 at 01:58:10PM +0800, ma qiang wrote: > > > >> Hi all, > > > >> I have install xen 3.4.3 based on centos5.4. > > > >> I changed a line in /etc/xen/xend-config.sxp like > below: > > > >> (network-script network-bridge) ----> (network-script > > > my-network-bridge) > > > >> > > > >> and #cat /etc/xen/scripts/my-network-script prints as > below: > > > >> #!/bin/sh > > > >> /etc/xen/scripts/network-bridge "$@" netdev=eth0 > bridge=eth0 > > > >> /etc/xen/scripts/network-bridge "$@" netdev=eth1 > bridge=eth1 > > > >> > > > >> Now I installed two vms, and set one vm using eth0, then > set > > the other > > > >> using eth1. > > > >> So, I can connect internet in both vms. > > > >> > > > >> But now my problem is that the throughput of two network > card > > is the > > > >> same as only one network card. > > > >> Another words, If I use the only one network card such as > eth0 > > for the > > > >> two vms, the throughput of eth0 is about 250m. > > > >> But when I set one vm use eth0 and the other use eth1, the > > total > > > >> throughput of eth0 and eth1 is about 250m too. > > > >> > > > >> That why? Any bottleneck? > > > >> Thanks a lot. > > > >> > > > > > > > > Do you mean 250 Mbit/sec ? > > > > > > > > You should get much more.. how are you measuring the > > performance? > > > > What benchmark are you using? > > > > > > > > Are both the eth0 and eth1 connected to the same > network/vlan? > > > > > > > > When you run the network benchmark try running "xm top" > > > > to monitor the overall cpu usage.. also run "top" in dom0 > > > > to see if there''s a big cpu usage in dom0. > > > > > > > > -- Pasi > > > > > > > > > > > > > > References > > > > > > Visible links > > > 1. mailto:[4]pasik@iki.fi > > > > References > > > > Visible links > > 1. mailto:maqiang1984@gmail.com > > 2. mailto:pasik@iki.fi > > 3. mailto:pasik@iki.fi > > 4. mailto:pasik@iki.fi >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Fajar A. Nugraha
2010-Sep-06 07:12 UTC
Re: [Xen-users] The bottleneck of using multinetwork card?
On Mon, Sep 6, 2010 at 1:57 PM, ma qiang <maqiang1984@gmail.com> wrote:> 20% is the vm''s utilization of cpu, not dom0''s. > My domU is hvm and centos linux.HVM is known to have slower disk/network I/O. Here are some things you could check: - look at CPU performance on both dom0 and domU using a program that can show individual CPU load (like htop). My guess is one CPU is at 100% utilization - check to see if you''re already using PV drivers. If not, using it might increase throughput. See http://pastebin.com/fb6fe631 for example (basically you need to make sure the correct modules are loaded while others are blacklisted) - Try using PV domU, and compare the results. -- Fajar _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Pasi Kärkkäinen
2010-Sep-06 07:19 UTC
Re: [Xen-users] The bottleneck of using multinetwork card?
On Mon, Sep 06, 2010 at 03:09:47PM +0800, ma qiang wrote:> >On Mon, Sep 06, 2010 at 02:52:16PM +0800, ma qiang wrote: > > I did the same test one the same computer in dom0 using only one > network > > card, the results show 1000M/sec. > > > >So what kind of guests are you using the benchmark from? > That is IXIA side based on centos 5.4 also. >In an earlier email you said it was Xen HVM guest. You need to install PV-on-HVM drivers to it. It won''t be fast otherwise. Or then install it as a PV guest, that''ll be fast out-of-the-box. -- Pasi> On Mon, Sep 6, 2010 at 3:02 PM, Pasi Kärkkäinen <[1]pasik@iki.fi> wrote: > > On Mon, Sep 06, 2010 at 02:58:32PM +0800, ma qiang wrote: > > Sorry, I lost two words. > > I reboot in bare linux not dom0, and have tested it on the same > computer > > using only one network card. the results show 1000M. > > > > So what do you get from xen dom0? > In dom0 and bare linux, the results show both 1000M. But in domU, the > results are bad. > > > > Also if your guest is HVM, have you installed PV-on-HVM drivers? > without those drivers the performance will be bad from the guest. > PV-on-HVM drivers ? Sorry, I haven''t installed that. And I''m going > to search that. > > > > -- Pasi > > On Mon, Sep 6, 2010 at 2:57 PM, ma qiang > <[1][2]maqiang1984@gmail.com> wrote: > > > > 20% is the vm''s utilization of cpu, not dom0''s. > > My domU is hvm and centos linux. > > "test server" is OK, and I have tested is on the same computer > using > > only one network card. the results show 1000M. > > > > On Mon, Sep 6, 2010 at 2:50 PM, Pasi Kärkkäinen > <[2][3]pasik@iki.fi> wrote: > > > > On Mon, Sep 06, 2010 at 02:44:59PM +0800, ma qiang wrote: > > > > Do you mean 250 Mbit/sec ? > > > Yes. > > > > You should get much more.. how are you measuring the > > performance? > > > > What benchmark are you using? > > > I use an tool from ixia to test the network in the vm. > > > > Try using some standard network performance testing tools > > such as "iperf" or just regular ftp file transfers. > > > > Are both the eth0 and eth1 connected to the same > network/vlan? > > > Yes. I run my test from another computer to connect to > vm0 and > > vm1, and > > > they all in the same switch. > > > To be noted my cards are both 1000m network card. > > > > > > > How is the ''test server'' connected? > > > > > > When you run the network benchmark try running "xm top" > > > > to monitor the overall cpu usage.. also run "top" in > dom0 > > > > to see if there''s a big cpu usage in dom0. > > > No, the cpu is less than 20% > > > > 20% on dom0? > > How about the domUs? Do they have big cpu usage? > > > > What kind of guests do you have? pv? hvm? linux? windows? > > > > -- Pasi > > > On Mon, Sep 6, 2010 at 2:08 PM, Pasi Kärkkäinen > > <[1][3][4]pasik@iki.fi> wrote: > > > > On Mon, Sep 06, 2010 at 01:58:10PM +0800, ma qiang > wrote: > > > >> Hi all, > > > >> I have install xen 3.4.3 based on centos5.4. > > > >> I changed a line in /etc/xen/xend-config.sxp like > below: > > > >> (network-script network-bridge) ----> > (network-script > > > my-network-bridge) > > > >> > > > >> and #cat /etc/xen/scripts/my-network-script prints as > below: > > > >> #!/bin/sh > > > >> /etc/xen/scripts/network-bridge "$@" netdev=eth0 > bridge=eth0 > > > >> /etc/xen/scripts/network-bridge "$@" netdev=eth1 > bridge=eth1 > > > >> > > > >> Now I installed two vms, and set one vm using eth0, > then set > > the other > > > >> using eth1. > > > >> So, I can connect internet in both vms. > > > >> > > > >> But now my problem is that the throughput of two > network card > > is the > > > >> same as only one network card. > > > >> Another words, If I use the only one network card such > as eth0 > > for the > > > >> two vms, the throughput of eth0 is about 250m. > > > >> But when I set one vm use eth0 and the other use eth1, > the > > total > > > >> throughput of eth0 and eth1 is about 250m too. > > > >> > > > >> That why? Any bottleneck? > > > >> Thanks a lot. > > > >> > > > > > > > > Do you mean 250 Mbit/sec ? > > > > > > > > You should get much more.. how are you measuring the > > performance? > > > > What benchmark are you using? > > > > > > > > Are both the eth0 and eth1 connected to the same > network/vlan? > > > > > > > > When you run the network benchmark try running "xm top" > > > > to monitor the overall cpu usage.. also run "top" in > dom0 > > > > to see if there''s a big cpu usage in dom0. > > > > > > > > -- Pasi > > > > > > > > > > > > > > References > > > > > > Visible links > > > 1. mailto:[4][5]pasik@iki.fi > > > > References > > > > Visible links > > 1. mailto:[6]maqiang1984@gmail.com > > 2. mailto:[7]pasik@iki.fi > > 3. mailto:[8]pasik@iki.fi > > 4. mailto:[9]pasik@iki.fi > > References > > Visible links > 1. mailto:pasik@iki.fi > 2. mailto:maqiang1984@gmail.com > 3. mailto:pasik@iki.fi > 4. mailto:pasik@iki.fi > 5. mailto:pasik@iki.fi > 6. mailto:maqiang1984@gmail.com > 7. mailto:pasik@iki.fi > 8. mailto:pasik@iki.fi > 9. mailto:pasik@iki.fi_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On Mon, Sep 6, 2010 at 3:12 PM, Fajar A. Nugraha <fajar@fajar.net> wrote:> On Mon, Sep 6, 2010 at 1:57 PM, ma qiang <maqiang1984@gmail.com> wrote: > > 20% is the vm''s utilization of cpu, not dom0''s. > > My domU is hvm and centos linux. > > HVM is known to have slower disk/network I/O. > Here are some things you could check: > - look at CPU performance on both dom0 and domU using a program that > can show individual CPU load (like htop). My guess is one CPU is at > 100% utilization >*I''m sure none CPU is at 100% utilization. I double confirmed that. :)*> - check to see if you''re already using PV drivers. If not, using it > might increase throughput. See http://pastebin.com/fb6fe631 for > example (basically you need to make sure the correct modules are > loaded while others are blacklisted) > - Try using PV domU, and compare the results. > *I don''t know if I have already using PV drivers, I''m going to check it. > Thanks a lot.* > -- > Fajar >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
I have tried the methods as the content of this link. But when I trye: # [root@test boot]# mkinitrd -v initrd-2.6.18-194.11.1.el5xen.img 2.6.18-194.11.1.el5xen prints as belows: Creating initramfs Looking for deps of module ehci-hcd Looking for deps of module ohci-hcd Looking for deps of module uhci-hcd Looking for deps of module ext3: jbd Looking for deps of module jbd Looking for driver for device sda2 Looking for deps of module pci:v00001000d00000060sv00001028sd00001F0Cbc01sc04i00: scsi_mod megaraid_sas Looking for deps of module scsi_mod Looking for deps of module sd_mod: scsi_mod Looking for deps of module megaraid_sas: scsi_mod Looking for deps of module pci:v000010B5d00008518sv00000000sd00000000bc06sc04i00: shpchp Looking for deps of module shpchp Looking for deps of module pci:v000010B5d00008518sv00000000sd00000000bc06sc04i00: shpchp Looking for deps of module pci:v00001166d00000144sv00000000sd00000000bc06sc04i00: shpchp Looking for deps of module sata_svw: scsi_mod libata Looking for deps of module libata: scsi_mod Looking for deps of module xen-vbd No module xen-vbd found for kernel 2.6.18-194.11.1.el5xen, aborting. That''s why? I have yum installed xen 3.4.3 from http://www.gitco.de/repo/xen3.4.3/, and based on centos 5.4. Best Wishes! On Mon, Sep 6, 2010 at 3:26 PM, ma qiang <maqiang1984@gmail.com> wrote:> > > On Mon, Sep 6, 2010 at 3:12 PM, Fajar A. Nugraha <fajar@fajar.net> wrote: > >> On Mon, Sep 6, 2010 at 1:57 PM, ma qiang <maqiang1984@gmail.com> wrote: >> > 20% is the vm''s utilization of cpu, not dom0''s. >> > My domU is hvm and centos linux. >> >> HVM is known to have slower disk/network I/O. >> Here are some things you could check: >> - look at CPU performance on both dom0 and domU using a program that >> can show individual CPU load (like htop). My guess is one CPU is at >> 100% utilization >> > *I''m sure none CPU is at 100% utilization. I double confirmed that. :)* > > >> - check to see if you''re already using PV drivers. If not, using it >> might increase throughput. See http://pastebin.com/fb6fe631 for >> example (basically you need to make sure the correct modules are >> loaded while others are blacklisted) >> - Try using PV domU, and compare the results. >> *I don''t know if I have already using PV drivers, I''m going to check it. >> Thanks a lot.* >> -- >> Fajar >> > >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Fajar A. Nugraha
2010-Sep-06 13:31 UTC
Re: [Xen-users] The bottleneck of using multinetwork card?
On Mon, Sep 6, 2010 at 2:55 PM, ma qiang <maqiang1984@gmail.com> wrote:> I have tried the methods as the content of this link. > But when I trye: # [root@test boot]# mkinitrd -v > initrd-2.6.18-194.11.1.el5xen.img 2.6.18-194.11.1.el5xen prints as belows:That guide is supposed to be run on RHEL/Centos 5 HVM dom0. Not dom0. Not PV domU. You''re using kernel-xen RPM (note the el5xen ending?). Check the guide, and compare the versions with what you have. Are you sure you''re using HVM domU? -- Fajar _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Yes, I did these thing in dom0. Should I have to reboot to bare centos, and redo these steps? I use the kernel="/usr/lib/xen/boot/hvmloader" in my config file, and I have to enable cpu vt, or I can''t start domU successfully. Additionally, I can install windows and linux for domU. So I think I use the HVM domU. On Mon, Sep 6, 2010 at 9:31 PM, Fajar A. Nugraha <fajar@fajar.net> wrote:> On Mon, Sep 6, 2010 at 2:55 PM, ma qiang <maqiang1984@gmail.com> wrote: > > I have tried the methods as the content of this link. > > But when I trye: # [root@test boot]# mkinitrd -v > > initrd-2.6.18-194.11.1.el5xen.img 2.6.18-194.11.1.el5xen prints as > belows: > >> That guide is supposed to be run on RHEL/Centos 5 HVM dom0. Not dom0. > Not PV domU. > You''re using kernel-xen RPM (note the el5xen ending?). Check the > guide, and compare the versions with what you have. > >> Are you sure you''re using HVM domU? > > -- > Fajar >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Fajar A. Nugraha
2010-Sep-07 03:07 UTC
Re: [Xen-users] The bottleneck of using multinetwork card?
On Tue, Sep 7, 2010 at 8:09 AM, ma qiang <maqiang1984@gmail.com> wrote:> Yes, I did these thing in dom0. > Should I have to reboot to bare centos, and redo these steps?No. Do it on the HVM domU. You don''t need to modify anything in dom0 to activate PV drivers. ... also, just a reminder, since you modified the dom0 earlier, make sure you revert the changes (e.g. make sure the initrd on dom0 is the correct one, not the one you tried to create an failed). -- Fajar _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users