Latency between FEX and Nexus 9300

Does anyone know or tested latency from port to port on a Nexus FEX? We are seeing some weird/strange results to MTU above 1500. Everything has been configured for jumbo (9216). The FEX is a N2K-C2232PP-10GE and its attached to a Nexus9000 C9396PX 1/10G Module.
Port to Port Layer 2 on the FEX we are seeing around 76us
Traffic from the traffic generator connected to a FEX, to the traffic generator on the Nexus 9300, we are seeing around 74us
Traffic from the traffic generator connected to the Nexus 9300, to the traffic generator on the FEX, we are seeing around 2us.
Any ideas why the latency seems high in one direction? Same test on the 9300 only is fine also.

Ahmed,
Sorry for taking this long to get back to you.
Nexus 1000v is a virtualized switch and as such will require that any traffic coming in or leaving the VM will first need to pass through the virtualization layer, therefore causing a minimum delay that for some applications (VMs) can be catastrophic enough that may mean too much delay.
With VM-FEX you gain the option to bypass the virtualization layer with for example "Pass-Through" mode where the vmnics are really assigned and managed by the OS, minimizing the delay and making the VMs look as if they were directly attached, also, this offloads CPU workload in the mean time, optimizing the host/VM's performance.
The need for one or the other will be defined as always by the needs your organization/business has.
Benefits of VM-FEX (from cisco.com):
Simplified operations: Eliminates the need for a separate, virtual networking infrastructure
Improved network security: Contains VLAN proliferation
Optimized network utilization: Reduces broadcast domains
Enhanced application performance: Offloads virtual  machine switching from host CPU to parent switch application-specific  integrated circuits (ASICs)
Benefits of Nexus 1000v here on another post from Rob Burns:
https://supportforums.cisco.com/thread/2087541 
https://communities.vmware.com/thread/316542?tstart=0
I hope that helps 
-Kenny

Similar Messages

  • VM-FEX and Nexus 1000v relation

    Hi
    I am a new in virtulaization world and I need to know what is the relation between Cisco Nexus 1000v and Cisco VM-FEX?, and when to use VM-FEX and when to use Nexus 1000v.
    Regards

    Ahmed,
    Sorry for taking this long to get back to you.
    Nexus 1000v is a virtualized switch and as such will require that any traffic coming in or leaving the VM will first need to pass through the virtualization layer, therefore causing a minimum delay that for some applications (VMs) can be catastrophic enough that may mean too much delay.
    With VM-FEX you gain the option to bypass the virtualization layer with for example "Pass-Through" mode where the vmnics are really assigned and managed by the OS, minimizing the delay and making the VMs look as if they were directly attached, also, this offloads CPU workload in the mean time, optimizing the host/VM's performance.
    The need for one or the other will be defined as always by the needs your organization/business has.
    Benefits of VM-FEX (from cisco.com):
    Simplified operations: Eliminates the need for a separate, virtual networking infrastructure
    Improved network security: Contains VLAN proliferation
    Optimized network utilization: Reduces broadcast domains
    Enhanced application performance: Offloads virtual  machine switching from host CPU to parent switch application-specific  integrated circuits (ASICs)
    Benefits of Nexus 1000v here on another post from Rob Burns:
    https://supportforums.cisco.com/thread/2087541 
    https://communities.vmware.com/thread/316542?tstart=0
    I hope that helps 
    -Kenny

  • High latency between session and session step !

    Hi everyone,
    one Operator's picture link to describe :
    http://www.glowfoto.com/static_image/24-025726L/3954/png/09/2010/img6/glowfoto
    When you look at session 18 INT_520_T_DWH_ME...., it begins at 11:28:10. Then when you look at it's first step 65 it starts at 11:29:18.
    Do you have any idea why we get some much latency between the session and the first session step ? Almost 70 seconds waiting (or doing something else maybe ?)
    Thanks in advance
    Pierre-Henri

    hi there was a problem in the Manchester area earlier today but that is now resolved
    the first hop to your home hub is very high typically <1 not 305 may be worth trying to check if your network card drivers are up to date also try if you can another cable also try another port on your hub just as some extra suggestions
    If you want to say thanks for a helpful answer,please click on the Ratings star on the left-hand side If the reply answers your question then please mark as ’Mark as Accepted Solution’

  • VPC between server and nexus switches

    Hi all,
    I would like to ask following question:
    We are using virtual portchannels to connect e.g. ESXi-Servers with trunks on two Nexus switches. When the switch "rz2sw1" has gone faulty, we expect that the trunk connections between switch rz2sw2  would still work. That's the reason why they are built. But they didn't as you can see in the screenshot (attaced) former mail.
    Do you know what can be problem.. thank you

    It appears that the peer-link was down.  If the peer-link is down and the keep-alive is still up, the secondary device will shutdown all of it's VPC links to ensure dual active (split brain) is avoided.
    On a Nexus 7000, its recommended to use a front panel (data port) for the keep-alive instead of the management port.  My assumption would be that you are using the MGMT0 for keep-alive and all front panel ports went off-line.  If you use a front panel port for keep-alive and all line cards go off line on the primary, the secondary will take over as expected.

  • DCBX Between Linux and Nexus QOS

    Hi Guys
    i have a server with 2 physical links only.
    over this link im passing 2 vlans in trunk ( can be Portchannel to VPC or bond active passive , doesnt matter to me ) .
    vlan 302 is for data
    vlan 303 is for backup
    server on vlan 303 is pushing traffic on that link and would take as much
    bandwidth as we would give it , and would therefore cause communication issues to vlan 302 . ( congestion ) .
    I would like to talk about my options here , to avoid congestions .
    thought about enabling dcbx on clients ( linux ) and pfc ( priority flow
    control ) to be able to negotiate qos and bw allocation according to COS IEEE 802.1p .
    the idea is to have either the traffic prioritized/limit/shaped before/when its ingressing the nexus port .
    the dcbx looks very cool like i said , it's kind of like what the nexus has with the CNA Card .
    when it can send the linux nic through dcbx the qos parameters to apply ! ( ah ok , i see now its exactly the same )
    ps linux dcbx link : http://manpages.ubuntu.com/manpages/precise/man8/dcbtool.8.html
    Ingress Queuing Policies
    You can associate an ingress policy map with an Ethernet interface to guarantee bandwidth for the specified traffic class or to specify a priority queue.
    The ingress policy is applied in the adapter to all outgoing traffic that matches the specified CoS value.
    When you configure an ingress policy for an interface, the switch sends the configuration data to the adapter. If the adapter does not support the DCBX protocol or the ingress policy type-length-value (TLV), the ingress policy configuration is ignored.
    did anyone here tried it or implemented dcbx before ?
    anyway id be happy to hear your wise and helpfull suggestions please
    Thanks Thanks !!!

    Found the answer.
    Anatomy of RX packet counters on a Cisco Nexus 5548:
    Nexus5548# show int eth1/29 
      RX      
        6387682660 unicast packets  1495485 multicast packets  164 broadcast packets 
        6389178309 input packets  589693485138 bytes
        5146969 jumbo packets  0 storm suppression bytes
      * "input" = unicast + multicast + broadcast (all non-jumbo)
      * "jumbo" = frames sized 1519 bytes to the MTU bytes defined on the interface
      * "jumbo" frames are counted separately from non-jumbo ("input")
      * (unicast, multicast and broadcast "jumbo" frames appear to be rolled into the 
            one "jumbo" counter)
    This can be seen another way with the 'count detail' command, correlated to "show int" output:
    Nexus5548# show int eth1/29 count detailed
    Ethernet1/29
      Rx Packets:                                  6389178309 <-- input
      Rx Unicast Packets:                          6387682660 <-- unicast
      Rx Multicast Packets:                           1495485 <-- multicast
      Rx Broadcast Packets:                               164 <-- broadcast
      Rx Jumbo Packets:                               5146969 <-- jumbo
      Rx Bytes:                                  589693485138
      Rx Packets from 0 to 64 bytes:                  1059418 <-- non-jumbo start
      Rx Packets from 65 to 127 bytes:             6072007478       +
      Rx Packets from 128 to 255 bytes:              11055520       +
      Rx Packets from 256 to 511 bytes:             287107435       +
      Rx Packets from 512 to 1023 bytes:             12801466       +
      Rx Packets from 1024 to 1518 bytes:                  23 <-- non-jumbo end
      Rx Trunk Packets:                            6387682823 <-- unicast + broadcast
    fyi.

  • Latency between audio and keyboard commands (10.9 and 10.10)

    Apologies for the length of this. Don't pay too much attention to the precise details of my setup, because we're dealing with practically identical systems, where the problem occurs depending on the version of OSX.
    I'm on a 2010 Mac Pro, 2.66GHz 12-core, 32GB RAM. Audio via Firewire into a MOTU UltraLite mk1
    Basically, I use a program called Transcribe!, which is for analysing music. You can slow down audio files, loop sections, etc, but most importantly it displays a waveform where you can add markers to show where the beats, measures and sections are (for easier navigation). You do this by tapping on the keyboard... B to enter a beat, etc.
    In both Mavericks and Yosemite, something is going wrong.
    If I simply play the file, the waveform scrolls by smoothly, and I can see that the "current position" indicator is accurately reflecting what I'm hearing (a big peak for a cymbal crash, etc). But if I then start tapping the beats on the keyboard, the waveform starts to scroll jerkily and the beat markers land about 130 milliseconds late... that's pretty disastrous when you're trying to accurately analyse the music.
    Obviously a lot of variables, but here's the weird thing. THIS WORKS PERFECTLY IN MOUNTAIN LION (on the same computer). Everything is on the same computer, with the same peripherals. The only difference is the OSX version and the hard drive it's installed on... but I've tried this on two different drives (one SSD, one HDD) and with both 10.9 and 10.10.
    (And another weird thing - I can't recreate the problem on my MBPr. It seems to be restricted to 10.9 and 10.10 on this Mac Pro)
    I can't see anything meaningful in Console or Activity Monitor, and my basic System Preferences are all the same. The actual processing of audio and video is fine... this seems to be some kind of issue with keyboard control. Any suggestions gratefully received.

    A couple of little possibilities were bugging me.
    Works fine on MBP, but not on Mac Pro... could be a dodgy keyboard? Nope, plugged same keyboard into MBP and it's fine.
    Could be an actual fault on this particular Mac Pro, that somehow only manifests itself in Mavericks/Yosemite? Nope, plugged in my older (2008) Mac Pro and the same thing happened. Latency with 10.9 and 10.10, but not on earlier versions of OSX.
    Really don't get this!

  • Data Guard Archive Log Latency Between Primary and Physical Standby

    How can I get the time it takes (latency) for the primary instance to get an archive log over to the physical standby instance and get it "archived" and "applied". I have been looking at the V$ARCHIVED_LOG view on each side but the COLUMN "COMPLETION_TIME" always shows a date "MM/DD/YY" and no timestamp. I thought the DATE datatype include data and time. Any ideas on how I can get the latency info I'm looking for?
    Thanks
    Steve

    the COLUMN "COMPLETION_TIME" always shows a date "MM/DD/YY"
    and no timestamp. Did you try using TO_CHAR ? e.g.
    to_char(completion_time,'dd/mm/yyyy hh24:mi:ss')

  • Connection 10 gig between 6500 and Nexus 7009

    I am not able to get link between a 10 gig interface with a 10gbase-lrm on a nexus 7k trying to connect to a 6500 with a 10 gig 10gbase-lx4 interface. I was told this would work. Anyone know if the interfaces are compatable. If not what interface do i need to get for the nexus 7k box.
    Thank You.                   

    Please check mentioned links. I hope it will help.
    http://www.cisco.com/en/US/prod/collateral/modules/ps2797/ps5138/product_data_sheet09186a008007cd00.pd
    http://www.cisco.com/en/US/prod/collateral/modules/ps5455/eol_c51_599855.html

  • LACP port channel between 6509 and Nexus 7K

    We are in the process of migrating from dual 6509's to dual 7010's.  We have moved our 5k/2K's behind the 7K and have layer 2 up between the 6509 and 7K.  This link is configured as a port channel with 2 1gig links using LACP.  The port channel is up and working and traffic is passing but it doesn't appear the load it equally distributed between the links.  Both the 7K and 6K are setup for src-dst-ip for the load balancing.  The links have been in place for over 12 hours and I would have expected them to "equal" out.  Has anyone had this issue in or is this to be expected?  For clarification there is not VPC inolved in this configuration it is simply a port-channel between one 6509 and a 7010.
    Thanks,
    Joe

    We are in the process of migrating from dual 6509's to dual 7010's.  We have moved our 5k/2K's behind the 7K and have layer 2 up between the 6509 and 7K.  This link is configured as a port channel with 2 1gig links using LACP.  The port channel is up and working and traffic is passing but it doesn't appear the load it equally distributed between the links.  Both the 7K and 6K are setup for src-dst-ip for the load balancing.  The links have been in place for over 12 hours and I would have expected them to "equal" out.  Has anyone had this issue in or is this to be expected?  For clarification there is not VPC inolved in this configuration it is simply a port-channel between one 6509 and a 7010.
    Thanks,
    Joe

  • Difference in latency between USB and FW interfaces?

    So I bought Logic Studio the other day, after thinking about it for a while. Seems great so far!
    I am planning to use Mainstage for several instruments, vocals etc., and I am about to go buy an audio interface. Originally thinking about the m-audio ProFire 2626 (Firewire), I discovered that the Fast Track Ultra 8R (USB) has more (but probably lesser quality) Mic-preamps / instrument inputs out of the box. So, I'm wondering, will I experience more latency within Mainstage with a USB-interface compared to a FW-interface?

    best to try several. For Midi I use Amt8's with my Mini and G5, and go direct from
    a USB kybd controller (M Audio Keystation 88) with my Powerbook. Never had latency
    except with an Amt4 and the Powerbook. For audio, I use a Firepod fp10 or a Metric Halo2882.
    Never had latency issues. Never tried Midi and Audio on the same device...

  • High Latency between computer and home hub

    My internet began to run very slow last night with disconnects lasting about a minute. I have used BT for many years with very few problems.
    I ran a speed test through the bt site and it reports my line is working correctly.
    I ran a Tracert as follows
    C:\Users\Neil>tracert www.bbc.co.uk
    Tracing route to www.bbc.net.uk [212.58.246.92]
    over a maximum of 30 hops:
    1 366 ms 396 ms 305 ms BThomehub.home [192.168.1.254]
    2 292 ms 296 ms 335 ms 217.32.143.226
    3 366 ms 285 ms 292 ms 217.32.144.14
    4 287 ms 293 ms 287 ms 213.120.181.26
    5 290 ms 379 ms 306 ms 217.41.169.183
    6 328 ms 296 ms 287 ms 217.41.169.107
    7 332 ms 333 ms 289 ms acc1-10GigE-10-3-0.sf.21cn-ipp.bt.net [109.159.2
    51.97]
    8 302 ms 309 ms 301 ms core1-te0-0-0-7.ealing.ukcore.bt.net [109.159.25
    1.31]
    9 323 ms 288 ms 290 ms peer2-xe1-0-0.telehouse.ukcore.bt.net [109.159.2
    54.104]
    10 301 ms 297 ms 369 ms 194.74.65.42
    11 * * * Request timed out.
    12 287 ms 299 ms 374 ms ae0.er01.cwwtf.bbc.co.uk [132.185.254.93]
    13 295 ms 308 ms 304 ms 132.185.255.165
    14 308 ms 300 ms 298 ms bbc-vip013.cwwtf.bbc.co.uk [212.58.246.92]
    Trace complete.
    I have tested in on second computer with similiar results. Both are wired connections with no previous connection problems.
    Does this suggest a problem with the router/filter itself or could it also be a fault on BTs end?

    hi there was a problem in the Manchester area earlier today but that is now resolved
    the first hop to your home hub is very high typically <1 not 305 may be worth trying to check if your network card drivers are up to date also try if you can another cable also try another port on your hub just as some extra suggestions
    If you want to say thanks for a helpful answer,please click on the Ratings star on the left-hand side If the reply answers your question then please mark as ’Mark as Accepted Solution’

  • Connecting a Dell B22 FEX to Nexus 6001 and storage

    Hi Folks,
    We have a setup where we are running a Dell B22 FEX in Blade enviornment and want to connect the B22 FEX to a cisco Nexus 6001 switch. As per NX-OS release note, B22 FEX and 6001 Nexus connectivity is supported.
    Now after connecting to Nexus 6001, how do i get access to storage pool or SAN fabric ?  As per another thread of discussion, Nexus 6001 does not support direct fabric attachment at this point in time. So how do we bridge these two elements to a storage fabric ??
    As per 6.02 release note:
    "Support for DELL FEX
    Added support for the Cisco Nexus B22 Dell Fabric Extender for Cisco Nexus 6000 Series switches starting with the 6.0(2)N1(2) release."
    This is the exact reason we bought it. We have a enviornment where we are running Dell B22 FEX.  Idea is to connect the B22 FEX into Nexus 6001. We are confuse at this point. After connecting the Dell B22 FEX to Nexus 6001, how to access the storage network or storage fabric ??
    Thanks,

    Hi Rays,
    The function should be the same for all FEXs regardless of what parent switch they connect to. to.
    if you are referring to this comment:
    You can connect any edge switch that leverages a link redundancy mechanism not dependent on spanning tree such as Cisco FlexLink or vPC
    FelxLink is a different technology that does not use STP, but not every switch platform supports FelxLink. FlexLink is not used very often, as other technologies like VSS, VPC and stacking has emerged.
    HTH

  • What are different between software VM-FEX and hardware VM-FEX?

    Hi everybody,
    I am confuse software VM-FEX and hardware VM-FEX.
    Does the software VM-FEX is Emulated mode and hardware VM-FEX is VMdirectpath mode?
    If i design a VM-FEX, is there must have nexus 1000V or not ?
    Is  Nexus 1000V and VM-FEX must work together or select one of them?

    Have a look at http://keepingitclassless.net/2013/06/cisco-vm-fex-and-the-nexus-1000v/
    VM-FEX in hardware requires on the host side a VIC adaptor, and a UCS FI or N5k as a switch (and control plane)
    VM-FEX in hardware has 2 submodes (PTS = Emulated Mode) and UPS (path Through)
    VM-FEX makes use of the VEM module (same as N1K; it is operated in a different mode however)
    VEM can do N1k local switching or VM-FEX path through, but not both !

  • Can I configure FCoE between UCS Fiber Interconnect and Nexus 5k?

    Hello,
    I have 2 UCS with 2 pair of Fiber interconnect 6120XP and they are conencted to Nexus 5k. Currently 10Gig Ethernet and FC are used for connectvity. I have 8 port FC module in Nexus 5k. Storage is connected to Nexus 5k over FC. I would like to change the FI to Nexus 5K connectvity and configurtation  with FCoE so that I can get few more free FC ports for stoarge connecting. Could you please help me whether it is possible with Nexus OS version 5 and how do we do this?
    Thanks.

    What storage are you using and is it capable of running a CNA?
    If so, one configuration I have used to reduce ports out of the storage and reduce the need for a high density of FC ports is to run FCoE from Storage to 5k, then trad FC from 5k to FI.
    With the 5k acting as a bridge between FCoE and FC you don't hit the single-hop limitations that currently exist within the protocol.
    Sent from Cisco Technical Support iPhone App

  • Adapter-fex Supported VICs and Nexus Switches

    HI,
    I am thinking of using C220M4 in standalone mode. I have been reading about adapter-fex and I am wondering which Cisco VIC supports adapter-fex and which Nexus Switch will support adapter-fex.
    I am thinking of using a VIC1225 and Nexus 3000.
    Thanks.

    I believe that adaptor fex is supported only with Nexus 5k and N2k:
    Network Adapter Virtualization Design (Adapter-FEX) with Cisco Nexus 5500 Switches and Cisco Nexus 2232 Fabric Extenders
    http://www.cisco.com/c/en/us/products/collateral/switches/nexus-5000-series-switches/guide_c07-690080.html
    http://www.cisco.com/c/en/us/products/collateral/switches/nexus-5000-series-switches/data_sheet_c78-657397.html

Maybe you are looking for