4948, Etherchannel 10GbE links

Similar Questions

• EtherChannel or fiber channel on 3508?

  When you aggregate several fibers together on G 3508 GBIC, you configure as an etherchannel or fiber channel? I have looked around online for hours, but can not find what it mentioned anywhere.

  Thanks in advance,

  Brian

  EtherChannel (or link aggregate, or whatever you prefer).

  Fibre channel is a data link protocol used in storage networks and a different protocol for ethernet.

  Sent by Cisco Support technique Android app

• FT peaks of the processor and latency

  What kind of latency should expect when running FT protected comments?

  I have a new 3-node cluster using servers DL380 G9 6.0 with the latest patch from April. Nothing runs on the then invited another cluster vCenter and 1 2012 windows server, I use for the FT test. When I activate FT on the client, it becomes slow and the comments CPU spikes when you do something, open the windows, moving around, etc. It's ping to meet when protected FT is usually is normally 50 ms or more and you can see it increase when the guest cpu spikes as a result of the opening of windows.

  Is this normal? I do not think that this kind of slow and latency would be acceptable for production applications, am I missing something?

  the guest is 2012 with 2 vCPU and 8 GB of memory

  Exchanges of Vmotion and FT have 2 network cards dedicated to each of them and active jumbo frames. When traffic monitoring network connections is not anywhere close to saturating the network. Use of the host is nearly 5% on all nodes.

  After trials with network cards 10 GbE for IP and bandwidth monitoring, that we determined that 10 GB is required even if the documentation indicates that it is only recommended. They changed the underlying technology that connects FT in vSphere 6.0 vLockstep to what is called "Fast pointing to the audit. Followed by the natachasery FT using esxtop on an old 5.5 system that uses the old technology (vLockstep) and then this same comments and using FT on 6.0 while monitoring with esxtop once again, we could see that it uses a lot more bandwidth to do the same.

  So the real question wasn't that something was misconfigured, but the documentation needs to read that 10 GB is 'necessary' and not 'recommended '.

  Working with VMware, we also discovered that they would not escalate the ticket to engineering at least use 10GBe links for feet and inches 6.0, it would make you assume that it is an obligation, not a recommendation. The documentation is misleading and I hope that this post helps someone in the future and that VMware updates the requirements documentation.

• Network host connectivity problem

  Hi all
  I have a bit of a strange an ESXi 5 U1 host connectivity problem. The host has a network card in the 10.0.0.0/24 subnet, and from there it can ping both devices but cannot ping to two other devices. To explain a little better:
  There are two hosts with identical hardware (HP ProLiant DL380 G6) both built with the same picture of ESXi 5 U1 from HP.
  There is an IP address on the switch (Cisco 3750 gigabit) management.
  There are two servers running Openfiler, which I'll use for ISCSI data banks.
  All these devices are in the same VLAN in the 10.0.0.0/24 subnet. host 1 is identical to the 2 host configuration (bar IP address and host name, etc.). Host can ping all devices on the subnet 1 and maybe ping by all devices on the seubnet. 2, the one with the issue of the host can ping and ping the Cisco FoxyCode and host 1, but cannot ping or ping by one of the two boxes openfiler.
  Research on the openfiler boxes I see missing entries in the arp (marked as incomplete) table. Host 2 he entered complete arp for the openfiler boxes. A trace of the wireshark package shows the openfilers sending arp requests but no response from host 2. The Cisco and host 1 to get the answers.
  I have patched the critical fixes host 2 until the last critical and no, applied the latest versions of firmware HP etc. Swapped, cables, tried different switch ports and even different NICs on host 2, all to nothing does not. The Cisco ports configured all identically from host 1. The openfiler boxes are set up with interfaces servile and the Cisco switch ports are etherchannelled for these. There is no etherchannel or link aggregation etc for guests, straight connection to the front in the VLAN.
  Interestingly, all devices also has an interface in the 192.168.2.0/sous-reseau 24 for management and all devices can see all other devices here absolutely fine.
  If anyone has any ideas on where to go from here please let me know, I'm going crazy trying to understand this one!

  ts999 wrote:

  The openfiler boxes are set up with interfaces servile and the Cisco switch ports are etherchannelled for these.

  I wonder if the Etherchannels in Openfiler servers might not work. Depending on how the switch load balances executives, this could cause a host to work and the other is not.

• iSCSI Multipathing - really useful?

  Suppose to have this configuration:

  2 ESXi host-> HP DL380 G7 (4 x 1 Gbit NIC + 2 x 10 Gbit NIC)

  1 storage-> MSA P2000 (2 x 1 Gbit NIC + 4 x 10 Gbit NIC)

  The two hosts will be connected to the storage through 10 Gbit adapter. Since no 10 Gbit switch is available, guests will be DIRECTLY connected to the storage.

  Multipathing is always necessary in this case? I mean, I can understand, multiple paths is very useful if you have several path to reach the storage (multiple switch with a network different for examples). But in this case, if for example a 10 Gbit card on one server will be lost, I would therefore lost two way.

  Am I wrong?

  Is 4 x 10 GbE NIC on storing a single NETWORK card? Otherwise, you can cross the cables from hosts to it and have a warranty from the point of view of the moribund NIC storage.

  Since I read your post tos ay that the cards 2 x 10 GbE NIC on the hosts are a single card, and if storage 4 x 10 GbE NIC is also a unique card, while the only reason for multipath would be to have access to the additional uplinks to a higher throughput of iSCSI.

  I would consider using the 1GbE NIC as an uplink ensures 10GbE links to help prevent single points of failure.

• The traffic is not correctly distributed through each link of etherchannel.

  Dear team,

  You can all have a clue on below.

  I have a cisco 3560 and created the port-channel switch and assign the port two to Po1 switch is connected to the HP VC.

  The question is highly uses a single port and on the other port no traffic is there.

  Please find below the config for your reference.

  Cisco 3560

  Switch1 #sh su etherchannel

  Flags: - Low P - D bundled in port-channel

  I have - autonomous s - suspended

  H Eve (LACP only)

  R - Layer 3 S - Layer2

  U - running f - cannot allocate an aggregator

  M - don't use, minimum contacts not satisfied

  u - unfit to tied selling

  w waiting to be aggregated

  d default port

  Number of channels in use: 4

  Number of aggregators: 4

  Protocol for the Port-Channel port group

  ------+-------------+-----------+-----------------------------------------------

  1 Po1 (SU) LACP Gi0/28 (P) Gi0/30 (P)

  2 Po2 (SU) - Gi0/22 (P) Gi0/23 (P) Gi0/24 (P)

  Gi0/25 (P) Gi0/26 (P) Gi0/27 (P)

  3 Po3 (SU) - Gi0/6 (P) Gi0/7 (P) Gi0/8 (P)

  Gi0/9 (P) Gi0/10 (P) Gi0/11 (P)

  4 (SU) Po4 - Gi0/1 (P) Gi0/14 (P)

  #sh Switch1 run int po1

  Building configuration...

  Current configuration: 166 bytes

  !

  Interface Port - Channel 1

  Description HP VC

  switchport trunk encapsulation dot1q

  switchport trunk allowed vlanXXXX

  switchport mode trunk

  end

  #sh Switch1 run int g0/28

  Building configuration...

  Current configuration: 223 bytes

  !

  interface GigabitEthernet0/28

  Server HP VC Description

  switchport trunk encapsulation dot1q

  switchport trunk allowed vlan XXXX

  switchport mode trunk

  channel-protocol lacp

  active in mode channel-group 1

  end

  #sh Switch1 run int g0/30

  Building configuration...

  Current configuration: 223 bytes

  !

  interface GigabitEthernet0/30

  Server HP VC Description

  switchport trunk encapsulation dot1q

  switchport trunk allowed vlan XXXX

  switchport mode trunk

  channel-protocol lacp

  active in mode channel-group 1

  end

  Switch1 #.

  Kind regards

  Prasad

  Disclaimer

  The author of this announcement offers the information in this publication without compensation and with the understanding of the reader that there is no implicit or explicit adequacy or adaptation to any purpose.  Information provided is for information purposes only and should not be interpreted as making the professional advice of any kind. Use information from this announcement is only at risk of the reader.

  RESPONSIBILITY

  Any author will be responsible for any damage that it (including, without limitation, damages for loss of use, data or profits) arising out of the use or inability to use the information in the view even if author has been advised of the possibility of such damages.

  Poster

   

  If you see a load imbalance to THE switch (i.e. from your server), then you should see if your server has any load balancing options like the switch has no control over this.

  If you see load imbalance OF switch (IE your server), then your need to select the best hash algorithm, for your traffic load balancing features.  Often by default of the switch is not the best.  Often using IP source and destination addresses works well.

  Sometimes, realize that the switch does not offer a good hash algorithm for your traffic for load balancing.  For example, if the majority of your traffic was between two hosts (replication, for example), all the traffic that would take the same link.  (Note: a hash of 6500 offers which includes port numbers for such situations, but not 3K switches.)

  This link http://www.cisco.com/c/en/us/support/docs/lan-switching/etherchannel/12023-4.html, explains your choices on different Cisco switches.

• EtherChannel does not support on 400 7600 - SIP

  Hello

  I'm a very big dilemma. I have 1 Gbps connection to my ISP. I'm under BGP to establish connectivity with the provider PE.

  Just to inform you, I have redundant connections on the connection of 2 - Cisco 3750 (main & secondary link) next to service provider and there are 4 routers for 2 services (primary and secondary each for 2 services) now use on services - 1A applications increases and reaches 900 ~ 950 MB/s, is not on the impact of the services-2 I pushed the 2 services via secondary link traffic. I have attached ppt for some knowledge

  Until a few weeks back the use was about 750 in the direction of entry and 700 Mbps in the exit direction of the WAN port on the service routers 1, but now the use of entry began to reach 900 ~ 950 MB/s in the direction of entry.

  I have 400 7600 - SIP (void crack-5XGE-SPA) on the routers [services for Qos ] are related to the switches. I spoke with the service provider, and they said that they would consider more and while I thought to do the first aggregation of ports-2 between the provider of Service & 3750 switch device and an another 2-port aggregation between 3750 Switch & services-1 (Cisco7606) router.

  There is no problem to create aggregation Service Provider betn ports device & switch 3750. It does not go between 3750 & services-1 switch (Cisco7606) router according to Cisco aggregation port document being not supported on 7600-SIP 400.

  It takes time for the service provider to provide 10 port of concert. I WS-X 6548-GE-TX module Services-(Cisco7606) 1 router. So, I thought WS-X 6548-GE-TX router connection ports on the router with switch and create port channel between Them.But I wonder how would function QOS can be used.

  So I need a Council if WS - X 6548-GE-TX ports support QOS with Etherchannel.

  Your advice will be appreciated

  R.

  Valerie

  The 3750 is a layer 3 switch.  Could you change the design to allow links to layer 3 through him?

  What get a 10Gbe for the 3750 and your 7600 module (s)?

• Network: Etherchannel vs. Fault Tolerance

  Here's my question.  I'm not a network guru but I understand if you assign NIC several ports to a virtual switch you gained transmit bandwidth (all ports are used to send), but that the same is not true on the receive side because you need a MAC and a layer of arbitration that can distribute traffic entering.  My hypothesis is that there is no way to go other than having two network adapters to the same switch and using Etherchannel trunking.

  So now, we have a trade off power.  If I choose alternative NIC to different switches for the fault tolerance, I am limited to only 1 GB on the side of the reception no matter what I do.

  If I have strange ESX hosts goes to switch A and even attending the B switch, it does not help much nor how many stores are building a reserve of 50% in their ESX capacity reservations.

  So assuming that my above assumptions are met, I'm curious to know what strategies and practices other stores allows to increase the pipeline of reception beyond 1 GB (without using 10gbE) without exposing to excessive exposure to a change of failure event.

  Thank you!

  Yes, that's basically it.

  I attended a VMWorld 2008 session given by a VMware employee who specializes in the robot of networking. He said that the "from port ID" is the method to use which is the reason why it is by default most recommended balancing. Hash TCP/IP method is OK, but has the configuration requirements of specific physical switch (EtherChannel), which makes it risky to be set up properly or poorly maintained. The hash of the IP would allow a simple VM for potentially communicate with multiple hosts on multiple physical links, but in most environments, it is not really necessary.

  Yes, when you have a redundant configuration of physical switches the redundant natachasery for the virtual switch should be distributed to multiple switches.

• Best wiring for blade M8024-k switch stack to ToR N3048 using 10GBe modules?

  

  I have a M1000e with 2 x M8024-k toggle (bunk) in B1 and B2 slots, I connected the switches rack via ports of 10GbE for the 2 x N3048 (bunk)

  What is the best way to connect?  And how do I add or number connections LAG as none of them are attached to the same switch?

  Thank you very much

  Battery

  Hello

  It is a good way to connect the two piles that you can configure the link as an aggregate of 4 X 10 giga links. You should use LACP rather than static, so that any broken link is quickly identified and thus will be abandoned in the aggregate.

• I have configured Link Aggregation on two Switch SF300

  The two links connect to both switches via two radios. The problem is that I can't access the attached radios http interface:

  radio_on_aggregation.PNG

  

  Everything else works fine and the radios are passing data according to PRTG monitoring of the towed gear

  Thank you

  Well, it looks like this is it.
  She looks like that ever, you can manage your bridges because of this.

  This is due to the notion of the Association and its operation
  L3 from the IP address of bridge management package can never be process as the etherchannel does not recognize. He treats only packets from his peers, etherchannel port in the etherchannel port.

  Overall, you cannot manage your bridges, to put it simply

• LACP. Recommendation of etherchannel, PaGP or static

  Hi all

  Some time ago I have red somewhere that the recommended way to create portchannels is to use the static method 'on' configuration. However, I am not able to find this information. Anyone here have a reference to the recommended method to configure portchannels?

  I found the following document:

  http://www.Cisco.com/en/us/docs/switches/LAN/catalyst6500/IOS/12.2Sx/best/practices/recommendations.html#wp1046803

  This document does not have an explicit statement that the method is preferable. Could someone help me with this please?

  Hi Michael,

  I would stay away from MODE in chanenling, a large part of the time that it leads to the switching loop.

  On the other hand, PAGP/LACP allows dynamically negotionation of etherchannel a being owners and the other open standard. LACP even allows active / standby group of several links and using some of them, but the others are in standby mode incase any member string fails.

  I'm not sure if it is registered anywere you can use on PAgp LACP or Pagp on LACP.

  If the grouping between the cisco devices, so we prefer to use the Pagp as is the owner of cisco.

  Kind regards

  Reem.

• How a loop shape in a misconfigured Etherchannel?

  Dear friends,

  It is a question frequently observed and practically proven that if both switches are interconnected by a number of parallel links that are grouped in an Etherchannel on a switch (of course, using the mode on ) while being unbundled on the second switch, a loop of Layer2 may well be created. However, I do not understand the exact mechanism of the formation of this loop.

  I am well aware of the basic principles behind: I know that STP treats interface Port-Channel as a single interface, and all Member links grouped in that Etherchannel share the same role/State PLEASE. I also understand that a frame of broadcast/multicast/unknown unicast sent by a port in the Etherchannel will reach the switch in front and cross flooded all the other links, arriving finally at the switch with the Etherchannel configuration.

  And right there, that is where my understanding ends: the framework came back and its destination is still unknown. However, from the point of view of the switch, the frame came in through a specific Port-Channel interface. If this switch flooding the frame, there he will flood through all ports except the port through which the framework was, which means that the frame will never sent through the Port-Channel. How the loop is created, then?

  Thank you so much for helping me with this!

  Best regards

  Peter

  Peter,

  The packages will be flooded with all interfaces for unknown broadcast/unicast packets with the exception of the interface when the flood came.

  In a 2 switch topology, switch A (which has the bundle) will flood the package on the members of the group waiting B switch to receive the package on its package.

  However, switch B will not receive the package on one of its physical interfaces and may experience flooding to B switch off its other physical interface connected which causes the STP loop.

  Kind regards

  Edison.

• Nexus5548up - is my etherchannel really 20G?

  Hi guys,.

  Im having trouble checking an etherchannel on my nexus and 6500...

  Now.

  I have a layer of distribution VSS (2 x 6500) and a bunch of nexus switches plugged.

  The link move as 1 x 10g to a switch and 1 x 10g to another in an etherchannel. (MEC)

  Even on the VSS, its an etherchannel.

  What I can't understand is if its really straight up to 20G because of the x 2 links in the GUY.

  An int see the watch always 10G but it shows the members... With two members, it should be a 20G GUY agreggated...

  I had put my statement of BS to 20000, but I don't think it really matters anyway...

  It goes the same for the end of 6500. Its a 10G UP UP channel ports. No mention of 20G.

  How can I be sure that it is infact running 20G 100%

  Thank you

  G

  port-channel 1 is on the rise

  Material: Port Channel, address: 547f.ee8c.9ae8 (bia 547f.ee8c.9ae8)

  Description: MEC01

  MTU 1500 bytes, BW 20000 Kbit, DLY 10 usec

  reliability 255/255, txload 1/255, rxload 1/255

  Encapsulation ARPA

  Port mode is trunk

  full-duplex, 10 Gb/s

  Input stream control is turned off, output flow control is disabled

  Switchport monitor is off

  EtherType is 0 x 8100

  Members in this channel: Eth1/1, Eth1/2

  Final cleaning of "show interface" counters never

  entry 30 seconds 4824 bps, 3 packets/s rate

  exit rate of 30 seconds 4296 bps, 3 packets/s

  Load-time #2: 5 minutes (300 seconds)

  Entry rate 2.03 Kbps, 2 pps; 960 output bit rate bps, 0 pps

  RX

  43 28 packets 1 multicasts packets broadcast unicast packets

  72 packages of entry 8909 bytes

  0 bytes removing storm Jumbo packets 0

  Giants of Runts 0 0 0 CRC 0 no buffer

  short entry error 0 0 0 overflow under-run 0 0 ignored

  0 0 bad etype watchdog fall 0 bad proto place 0 if drop

  0 entry with dribble 0 rejection of entry

  Rx pause 0

  TX

  30 packs 24 packages of multicast 3 broadcast unicast packets

  packages of exit 57 9612 bytes

  0 jumbo packets

  collision of errors 0 0 released 0 postponed end collision 0

  0 lost carrier 0 no 0 0 output babble carrier throw

  Tx break 0

  3 interface resets

  -CFG

  interface port-channel 1

  Description MEC01

  switchport mode trunk

  switchport trunk allowed vlan 1

  full duplex

  bandwidth 20000

  interface Ethernet1/1

  Description MEC01

  switchport mode trunk

  switchport trunk allowed vlan 1

  automatic speed

  full duplex

  channel-group 1

  interface Ethernet1/2

  Description MEC01

  switchport mode trunk

  switchport trunk allowed vlan 1

  automatic speed

  full duplex

  channel-group 1

  A given channel no matter what unique conversation will only flow in a single interface of the channel so your total thruput for a given conversation will be 10 gig.  You have 20 gig of pipeline to push all traffic as a part of the traffic goes down a single interface and some will go to the bottom of the other interface according to arr. algorithim you choose.

• WLC 4402 redundant link

  We have a WLC, currently it has a trunk single link in a 6506 and everything works fine.

  We've recently updated this network to the VSS system, so now we have two 6506, I want to connect the redundant port 4402 on the new switch, is there something special I need to do?

  as is a virtual switch that it should connect, does anyone know if WLC is capable of etherchannel or something similar?

  Thank you very much

  WLC called "LAG", but that's about the same. You can simply enable LAG or turn it off on the side of WLC. When activated, it will offer all of its ports. On the side of the switch that you need to configure "mode" on the portchannel, which means that the WLC is not compatible lacp or pagp.

  Nicolas

• Chassis & 10GbE blade M1000e through Modules

  We rotated M620 blades with 1 GB Pass Through modules without any problem. The M620s have the Qlogic\Broadcom cards mezzanines KR 10 GB for slots chassis A1\A2.

  We have just bought a few Pass through modules of KR 10GbE and exchanged out them. in the same locations of A1\A2. When the configuration of the BIOS, we get green link is turned on with our Cisco Nexus switch which shows a connection on the port, but when ESXi6 is loaded there is no link (orange light just on the module). NIC in ESX is set to automatic (will not force the 10 GB according to VMWare).

  Tested updates, driver updates, change cables Twinax Dell SFP + and fiber but no link. If I put the 1GbE modules back in they work fine.

  Strangely, at one point I did get network link in ESX running a single passthrough but have not been able to recreate it. Am I missing something that I do not see a reason for this which is why they should not work? This chassis we test on is quite old (2008) and we use just to test before deploy us in our new M1000e chassis. What is the material inside the frame at the origin of a problem that has subsequently been updated?

  You should be fair exchange in the chassis with the median plan 1.1.