Vmware vmxnet3 nic vs. e1000 vs. hardware-install - throughput performance



  • @yaxattax:

    I'm not going to conduct a formal test with no passthrough. I enabled passthrough on the WAN interface so that there was no possibility of traffic going anywhere except through pfSense. The LAN was a shared interface and I had already determined I wasn't going to get more than 100mbits throughput. As a result, I moved the LAN to having a dedicated device as well.

    Regarding power consumption, yes, pfSense has 1 CPU but the whole rig has 8 still. All cpu are controlled by the host (using linux, with better cpufreq drivers). As a result, idling in the VM uses less power than idling in baremetal pfSense.

    That's fine.. But it's impossible to conclude anything from your tests then.



  • I have a few questions:

    • Why run VLAN tagging with pfSense? Why not give leave the tagging up to ESX? This way pfSense passes everything untagged and ESX will tag it as it leaves. This method works fine for me, and prevents configuring the ESX for trunking.
    • Any benefit from running multiple physical NICs on the vSwitch? I'm running three 1-Gb NICs using "Route based on IP hash" load balancing, and I'm wondering if there is a benefit to running VMXNET3 or E1000. The VM still has one adapter per network though, so it's up to the ESX server to load balance and is transparent to the pfSense VM.


  • @FauxShow:

    I have a few questions:

    • Why run VLAN tagging with pfSense? Why not give leave the tagging up to ESX? This way pfSense passes everything untagged and ESX will tag it as it leaves. This method works fine for me, and prevents configuring the ESX for trunking.
    • Any benefit from running multiple physical NICs on the vSwitch? I'm running three 1-Gb NICs using "Route based on IP hash" load balancing, and I'm wondering if there is a benefit to running VMXNET3 or E1000. The VM still has one adapter per network though, so it's up to the ESX server to load balance and is transparent to the pfSense VM.

    1st question: For people with a single VM host (no clusters, or, at least, no v-motion) that idea is pretty much 6 of one 1/2 dozen of the other.  Probably doesn't matter much, do whichever you're more familiar with.  Personally, I'd do them at the ESX(i) host level as well, but feel free to do it either way.  Unless you have the situation you talk about next…

    2nd question:  I believe so.  For an ESX(i) host with multiple uplinks from its vSwitch, a VMXNET3 may help a lot.  A VMXNET3 is presented to the VM as a 10Gb adapter, as such, it'll pass more than 1Gb of traffic to a VM.  If you have multiple 1Gb connections you won't get more than 1Gb to any one destination, but you might be able to leverage more than 1Gb total to your VM, although I would not expect anywhere near a full 2Gb (mostly since the load balancing isn't based on load, so your two heavy destinations could easily end up on the same NIC.)

    Although, I do remember something about some vNICs being able to communicate at "bus" speed, ignoring their stated connection speed and transferring as fast as possible.  That was back in my VCP testing, so I don't exactly remember.  But, it would be easy to test.  Run 2 VM's on the same host, both with VMXNET3 vNICs, throw data around and see how fast it goes (generated data, though, files will be dependent on disk speed.)  Note CPU usage, though, remember, the vNICs are virtualized, they take CPU to run; if I recall correctly, this is the original idea behind the VMXNET NICs.

    You may notice I said "may help a lot" earlier.  I did hear somewhere that the "speed" of your vSwitch may be "set" by the fastest physical NIC connected to it.  Again, memory, not always as good as I'd like it to be; as well as my quick search google-fu.



  • @matguy:

    @FauxShow:

    I have a few questions:

    • Why run VLAN tagging with pfSense? Why not give leave the tagging up to ESX? This way pfSense passes everything untagged and ESX will tag it as it leaves. This method works fine for me, and prevents configuring the ESX for trunking.
    • Any benefit from running multiple physical NICs on the vSwitch? I'm running three 1-Gb NICs using "Route based on IP hash" load balancing, and I'm wondering if there is a benefit to running VMXNET3 or E1000. The VM still has one adapter per network though, so it's up to the ESX server to load balance and is transparent to the pfSense VM.

    1st question: For people with a single VM host (no clusters, or, at least, no v-motion) that idea is pretty much 6 of one 1/2 dozen of the other.  Probably doesn't matter much, do whichever you're more familiar with.  Personally, I'd do them at the ESX(i) host level as well, but feel free to do it either way.  Unless you have the situation you talk about next…

    2nd question:  I believe so.  For an ESX(i) host with multiple uplinks from its vSwitch, a VMXNET3 may help a lot.  A VMXNET3 is presented to the VM as a 10Gb adapter, as such, it'll pass more than 1Gb of traffic to a VM.  If you have multiple 1Gb connections you won't get more than 1Gb to any one destination, but you might be able to leverage more than 1Gb total to your VM, although I would not expect anywhere near a full 2Gb (mostly since the load balancing isn't based on load, so your two heavy destinations could easily end up on the same NIC.)

    Although, I do remember something about some vNICs being able to communicate at "bus" speed, ignoring their stated connection speed and transferring as fast as possible.  That was back in my VCP testing, so I don't exactly remember.  But, it would be easy to test.  Run 2 VM's on the same host, both with VMXNET3 vNICs, throw data around and see how fast it goes (generated data, though, files will be dependent on disk speed.)  Note CPU usage, though, remember, the vNICs are virtualized, they take CPU to run; if I recall correctly, this is the original idea behind the VMXNET NICs.

    You may notice I said "may help a lot" earlier.  I did hear somewhere that the "speed" of your vSwitch may be "set" by the fastest physical NIC connected to it.  Again, memory, not always as good as I'd like it to be; as well as my quick search google-fu.

    1. I think it's much easier to have ESX handle the VLAN tagging; it's one less thing to do on the VM to configure it properly, and it also prevents a user (authorized or not) from switching the network the VM is connected to from the VM.

    2. I think you are correct about the bus speed; I just ran iperf between two KNOPPIX VMs on different VLANs and had 1.6Gb/s using the E1000 adapter and 1.3Gb/s using the VMXNET3 adapter. This speed was also confirmed on the pfSense live Traffic Graph. It's odd that the E1000 adapters ran quicker than the VMXNET3 though, however the pfSense VM is also running E1000 adapters if that makes a difference. The pfSense VM has 8 CPUs (single virtual socket, eight virtual cores) and most were barely registering during that test; one hit 50% and another hit 100% briefly. The ESX server is a beast though, with dual Xeons running at 2.4GHz, each with 6 cores, plus hyper-threading for 24 logical processors.



  • @FauxShow:

    1. I think it's much easier to have ESX handle the VLAN tagging; it's one less thing to do on the VM to configure it properly, and it also prevents a user (authorized or not) from switching the network the VM is connected to from the VM.

    2. I think you are correct about the bus speed; I just ran iperf between two KNOPPIX VMs on different VLANs and had 1.6Gb/s using the E1000 adapter and 1.3Gb/s using the VMXNET3 adapter. This speed was also confirmed on the pfSense live Traffic Graph. It's odd that the E1000 adapters ran quicker than the VMXNET3 though, however the pfSense VM is also running E1000 adapters if that makes a difference. The pfSense VM has 8 CPUs (single virtual socket, eight virtual cores) and most were barely registering during that test; one hit 50% and another hit 100% briefly. The ESX server is a beast though, with dual Xeons running at 2.4GHz, each with 6 cores, plus hyper-threading for 24 logical processors.

    Do you notice any performance benefit from giving your pfSense VM so many cores?  I would imagine 2 or 3 being the max that pfSense can really utilize.  (Serious question, not saying you're doing anything wrong.)

    As for the transfer, I can imagine that going between like vNIC's on the same host possibly being faster than different.  Might be interesting to see VMXNET3 to VMXNET3.  (I don't have time to test today.)  For CPU usage, was that on the pfSense OS reporting CPU usage or VMware?


  • Banned

    Can you create a VLAN without a VLAN tag in Pfsense??



  • @Supermule:

    Can you create a VLAN without a VLAN tag in Pfsense??

    Are you looking to have mixed mode interface, with both a native, untagged network and tagged VLANs on the same interface?

    If this is in a VM, there really wouldn't be a need unless you really don't want to configure multiple vNIC interfaces in ESX(i).  If this is physical, remember, your switch will have to support VLANs anyway, so I'm not sure what benefit you'd be getting out of mixing the modes.

    It's a good academic question, and if I understood the quick look at the VLAN documents for pfSense, I think it does.  But, what are you trying to accomplish?



  • @matguy:

    Do you notice any performance benefit from giving your pfSense VM so many cores?  I would imagine 2 or 3 being the max that pfSense can really utilize.  (Serious question, not saying you're doing anything wrong.)

    As for the transfer, I can imagine that going between like vNIC's on the same host possibly being faster than different.  Might be interesting to see VMXNET3 to VMXNET3.  (I don't have time to test today.)  For CPU usage, was that on the pfSense OS reporting CPU usage or VMware?

    I've had stability issues with pfSense, so I'm making sure the VM has enough resources to keep running. I also switched back to 32-bit. Even if it can only fully utilize 2 there's no harm in giving 8, though if anyone can confirm a number I'd change it.

    I ran the test again between a VM running the E1000 adapter on one VLAN and iperf'ing to a user on another VLAN and was able to get 634Mb/s. So that's from the VM, through pfSense, to ESX, though a gigabit switch, and then to the user. Because I'm more concerned about stability than performance, and 634Mb/s is fine with me, I'm not going to switch to VMXNET2/3 adapters.



  • @FauxShow:

    @matguy:

    Do you notice any performance benefit from giving your pfSense VM so many cores?  I would imagine 2 or 3 being the max that pfSense can really utilize.  (Serious question, not saying you're doing anything wrong.)

    As for the transfer, I can imagine that going between like vNIC's on the same host possibly being faster than different.  Might be interesting to see VMXNET3 to VMXNET3.  (I don't have time to test today.)  For CPU usage, was that on the pfSense OS reporting CPU usage or VMware?

    I've had stability issues with pfSense, so I'm making sure the VM has enough resources to keep running. I also switched back to 32-bit. Even if it can only fully utilize 2 there's no harm in giving 8, though if anyone can confirm a number I'd change it.

    I ran the test again between a VM running the E1000 adapter on one VLAN and iperf'ing to a user on another VLAN and was able to get 634Mb/s. So that's from the VM, through pfSense, to ESX, though a gigabit switch, and then to the user. Because I'm more concerned about stability than performance, and 634Mb/s is fine with me, I'm not going to switch to VMXNET2/3 adapters.

    With a lot of cores per VM you can run in to scheduling issues of trying to schedule all the vCPUs at the same time.  Of course, with 24 cores at your disposal, this might not be an issue… yet.  If you start putting a lot of VMs (especially multi-core) on that host, watch your CPU Ready metrics, they'll tell you if you're having CPU scheduling issues.



  • @FauxShow:

    I have a few questions:

    • Why run VLAN tagging with pfSense? Why not give leave the tagging up to ESX? This way pfSense passes everything untagged and ESX will tag it as it leaves. This method works fine for me, and prevents configuring the ESX for trunking.
    • Any benefit from running multiple physical NICs on the vSwitch? I'm running three 1-Gb NICs using "Route based on IP hash" load balancing, and I'm wondering if there is a benefit to running VMXNET3 or E1000. The VM still has one adapter per network though, so it's up to the ESX server to load balance and is transparent to the pfSense VM.

    i'm letting pfsense handle the vlan tag because i have more vlans than it's possible to assign physical adapters.


  • Banned

    Yes but if you have DRS and failover switches(physical) you need the tag on PFsense, Vswitch and physical switch.

    I dont think you can run untagged VLAN's and migrate VM's to other cluster nodes if you dont do it this way.



  • @Supermule:

    Yes but if you have DRS and failover switches(physical) you need the tag on PFsense, Vswitch and physical switch.

    I dont think you can run untagged VLAN's and migrate VM's to other cluster nodes if you dont do it this way.

    Sure you can, all the labels across the port groups just need to be the same.  In reality, they don't even need to all be on the same switches, as long as the labels all match (for everyone's sanity, it's better if they're as close to exact mirrors as possible, though.)

    The only limitation that you "can't vMotion / DRS with, as far as network is concerned, is an internal only network.  If a VM is connected to a vSwitch that has no external physical NIC, it won't DRS it (it might let you manually vMotion it after a warning, I would have to test it with current versions.)

    I had a whole bunch of these I inherited at an old job, remnants of an um-restricted Lab Manager install for a DEV environment.  I just made each one their own VLAN and assigned them to port groups instead, that way they could vMotion till the cows came home and they didn't lose connection to each-other.  For performance reasons, I did create affinity rules to keep them together, though (they were groups of web, SQL, DC's in their little isolated networks.)


  • Banned

    I dont get that….

    How would you seperate traffic on the Vswitch if no VLAN tagging is done by Pfsense but only in Vsphere?

    How about the physical switch and vmotion across cluster nodes?

    You got me really confused here, since I spent a lot of time getting it to work so it could migrate VM's across nodes and more than one physical switch.



  • @Supermule:

    I dont get that….

    How would you seperate traffic on the Vswitch if no VLAN tagging is done by Pfsense but only in Vsphere?

    How about the physical switch and vmotion across cluster nodes?

    You got me really confused here, since I spent a lot of time getting it to work so it could migrate VM's across nodes and more than one physical switch.

    On a single vSwitch you create port groups, these port groups have a VLAN tag assigned to them and become "Networks" you can select in the vNIC settings for your particular VM.  Your VM will have as many vNIC's as you have port groups/networks that you need to connect pfSense to, one in each.  So, you can run in to the same issue that miloman has, where you run out of "virtual PCI slots" for your vNIC's, but if you only have a few VLANs, it works fine.

    I'm not saying it's a "better" way, just that it does work and can vMotion / DRS.

    Edit: note, a vNIC is seen in pfSense as its physical NIC(s).



  • @matguy:

    So, you can run in to the same issue that miloman has, where you run out of "virtual PCI slots" for your vNIC's, but if you only have a few VLANs, it works fine.

    BTW, the limit of Virtual NICs you can give a VM in ESXi 4 and up is 10 individual vNICs (up from 4 in ESX/ESXi 3.5.)


  • Banned

    Thats why you need tagging since pfsense is located in the "all" segment of the Vswitch and handles traffic to the individual VLAN's on the portgroups.

    I only have 2 vNIC's in PFsense and they are VLAN's in one, and none on the other interface.



  • @Supermule:

    Thats why you need tagging since pfsense is located in the "all" segment of the Vswitch and handles traffic to the individual VLAN's on the portgroups.

    I only have 2 vNIC's in PFsense and they are VLAN's in one, and none on the other interface.

    I hope we're using the same terminology in the same places.

    Anyone can set them either way and still have them vMotion-able as long as the labels your networks are connecting to and the underlying networks they're connecting to are the same.  (In fact, they'll vMotion even if the underlying networks are different, as long as they're labeled the same, whether it works after the vMotion or not is a different story.)

    Passing the VLANs through, rather than letting ESX(i) "sort" them in to individual vNICs is a matter of taste and comfort level (or, a matter of limitations if you have more than 8 or 9 networks to present to pfSense and start running out of vNIC "slots".)



  • Hello all,

    I have been doing my own testing comparison between vmxnet3 and e1000 running on ESXi 5.1 build 914609. The pfsense VM is configured with a single CPU and 1GB of RAM running on a dual socket Xeon X5675 (3.07Ghz) machine. These tests were done with installed 64-bit pfsense 2.0.2.

    I have a 10gige network so can test at speeds in excess of 1gige.

    I had to apply the tuning suggestions at http://fasterdata.es.net/host-tuning/freebsd to achieve top speeds.

    I had two main test scenarios:

    Scenario 1: iperf between the pfsense VMs and a linux VM (e1000) running on the same ESXi box connected to the same port group. This test does not hit a physical network.
    Scenario 2: iperf between the pfsense VM and an external linux machine connected via a 10gige switch and intel 10gige interface cards.

    For both of these scenarios I generated throughput with both e1000 and vmxnet3 three times and then took the highest value.

    Scenario 1:
    e1000: 2.42 Gbits/sec
    vmxnet3: 17.8 Gbits/sec

    Scenario 2:
    e1000: 2.8 Gbits/sec
    vmxnet3: 8.87 Gbits/sec

    So you can see that at greater than 1gbps speeds that vmxnet3 makes a huge difference. With inter-VM traffic on the host running 7 times faster than e1000.

    For additional information I ran speed tests between two CentOS 6.4 64-bit VMs with e1000 and achieved 26.1GBits/sec



  • @btbuilder:

    Hello all,

    I have been doing my own testing comparison between vmxnet3 and e1000 running on ESXi 5.1 build 914609. The pfsense VM is configured with a single CPU and 1GB of RAM running on a dual socket Xeon X5675 (3.07Ghz) machine. These tests were done with installed 64-bit pfsense 2.0.2.

    I have a 10gige network so can test at speeds in excess of 1gige.

    I had to apply the tuning suggestions at http://fasterdata.es.net/host-tuning/freebsd to achieve top speeds.

    I had two main test scenarios:

    Scenario 1: iperf between the pfsense VMs and a linux VM (e1000) running on the same ESXi box connected to the same port group. This test does not hit a physical network.
    Scenario 2: iperf between the pfsense VM and an external linux machine connected via a 10gige switch and intel 10gige interface cards.

    For both of these scenarios I generated throughput with both e1000 and vmxnet3 three times and then took the highest value.

    Scenario 1:
    e1000: 2.42 Gbits/sec
    vmxnet3: 17.8 Gbits/sec

    Scenario 2:
    e1000: 2.8 Gbits/sec
    vmxnet3: 8.87 Gbits/sec

    So you can see that at greater than 1gbps speeds that vmxnet3 makes a huge difference. With inter-VM traffic on the host running 7 times faster than e1000.

    For additional information I ran speed tests between two CentOS 6.4 64-bit VMs with e1000 and achieved 26.1GBits/sec

    Thanks for sharing this - any info on what CPU usage was like by the VM pfSense at these rates?



  • I'm afraid I wasn't looking. If I have to run these tests again I'll make a point of measuring.


Log in to reply