vswitch\pswitch and NIC load balancing

Similar Questions

• NIC teaming + with leaves HP VirtualConnect and Cisco load balancing

  Hello

  I configure ESX infrastructure running on the blades HP (c7000 enclosure), VirtualConnect and Cisco 3750 switches for the uplink. My network configuration is based on the HP VirtualConnnect Cookbook, scenario 11.

  On the ESX Server, I configured a vSwitch with two NICs grouped. Each card can see a different network of VirtualConnect. Each VC network has active SmartLink.

  Each network of VirtualConnect is associated with two ports on a Bay of VC. These two ports are the uplinks in Cisco 3750 switches. Cisco switches are configured in a pile, and each port associated with the VC network is configured with LACP.

  Here are the facts:

  -Each network (Host A and host B) is properly configured for the LACP Protocol, in other words, the links are shown as active/active in the VC Manager.

  -Communication of ESX is good.

  -Failover and failback function OK.

  The problem: I can't get the load balancer to work. All virtual machines use a single bear. I tried different algorithms on the server ESX (source port ID, mac, IP hash hash) and the configuration of the equivalent from Cisco.

  I have attached a diagram of the physical network, but also the setting to ports.

  What Miss me? Thank you very much

  Pablo

  Because you want to create technically not a channel in an ESX perspective, you do want to use the hash of the IP.  If ESX attempts to send traffic through the two network cards, the cisco switch should drop this package as part of its loop avoidance algorithms.  You should be able to check that out in the switch port statistics.  You should not use the switch port ID.

  Not sure about the promiscuous mode.  Who do not have any account, as far as load balancing is concerned, unless the portgroup was not properly inherit the vSwitch properties?

  -KjB

  VMware vExpert

• Vary the size of the host and the load balancing

  I have a 4 4-host ESX cluster that is almost the maximum ability to manage virtual machines ~ 80 (and growing).  The hosts are configured with 4 to 8 cores and 32 GB of RAM each.  RAM is currently the bottleneck; Processor is not far behind.  I plan to replace one of the hosts with a much larger machine, 12 cores, 96 GB of RAM.  If a host is significantly larger than the other hosts in a cluster, DRS will still do a good job of load balancing?  I guess the DRS will place multiple virtual machines on the host to the largest.  Is there a risk of DRS. putting too many virtual machines on the largest node?

  Until the cluster is in CVS mode, or processors are compatible vmotion, you can vary the size of the resource of physical machines in the cluster.  DRS will try to balance the load, the better it can over all cluster resources.  Depending on your level of DRS aggressive, improving to a particular machine in terms of past performance is considered, so more that likely, that you will have more vm on the host of the largest.

  -KjB

• Monitoring of Tuxedo 9.1 and RAC load balancing?

  Hi all
  
  I wrote this question in the old forum that BEA [http://forums.bea.com/thread.jspa?threadID=300006084] and now I have more question about monitoring in a tuxedo with RAC load balancing:
  
  How can I control in a tuxedo with RAC load balancing?
  How can I verify that the tuxedo load balancing is running correctly?
  How can I know that this group using my requests?
  
  We need to know if the query uses the instance ORARAC1 (groups of GRP_A1 and GRP_A2) or
  use failure ORARAC2 (groups of GRP_B1 and GRP_B2).
  
  There is a tool (same txrpt) that displays this information in a Tuxedo 9.1?
  
  Thanks in advance
  
  P.D.: Excuse my English

  Hello

  How can I control in a tuxedo with RAC load balancing?

  You can use the tmadmin pq and ESP see work queue of orders and the work done by each server.

  How can I verify that the tuxedo load balancing is running correctly?

  What do you mean by properly? This is an area where many clients think that Tuxedo isn't the load balancing, but in fact it is. In general, Tuxedo will select the first server offering a service that does all work queued to it. If all servers have jobs queued, then Tuxedo will select the server with the least amount of work waiting in line. Specifically, this means that if the system is not enough charge, the first server in the list of servers offering service will receive all applications. Note that this is what happens in a single machine and no data dependent routing and routing CARS in the picture. With data dependent routing or TUXRACGROUPS defined, the algorithm is similar, but the possible servers to consider are limited according to the setting of data dependent routing or TUXRACGROUPS as explained in the note on the other.

  How can I know that this group using my requests?

  Without TSAM (Tuxedo system) and Application Monitor, it will be nearly impossible to tell which groups and servers a particular request was sent to. What you trying to accomplish?

  We need to know if the query uses the instance ORARAC1 (groups of GRP_A1 and GRP_A2) or
  use failure ORARAC2 (groups of GRP_B1 and GRP_B2).

  Without changes to the application or using TSAM, it's going to be very hard to say. Why is it important for your application?

  Kind regards
  Todd little
  Chief Architect of Oracle Tuxedo

• Client based in 12 G load balancing

  We have a JDBC configuration against an address scan with client load balancing enabled.

  Example:

  JDBC:Oracle:Thin:@(Description=(LOAD_BALANCE=on)(Address=(Protocol=TCP)(Host=xxxxxxxxxxxx.de)(port=xxxx))(CONNECT_DATA=(service_name=XXXXXX)))

  What I understand of Diference between Client-side and Server Load Balancing If you do not use the server load balancing, you can bypass by connecting the service that identifies a particular node of RAC name.

  
  

  My question is, if you have a connection pool object this configuration and if the name service maps node to goes down, what happens to the connection connection pool?

  
  

  Issue 1) Don't scan address switches to the available according to the name of the service node, then the JDBC driver on the client must file all embusked connections and reconnect the available node? or is all the agnostic client failure of node on and all the old connections available in the pool are available for use?

  
  

  
  

  The client side or server balancing side basically works for the same purpose. The difference is that you do not have all the customers to reconfigure then change something in the environment, such as adding or removing a node such as the side server.

  Answer your question...

  If you mean the shared server architecture by connection pool, the shared server session connected to node 1 will die and everyone shares this session should log. Failover is not automatic, you must specify the failover clause in the JDBC or TNS connect string:

  TEST_TAF =

  (DESCRIPTION =

  (ADDRESS = (PROTOCOL = TCP)

  (HOST = rac - scan.example.com) (PORT = 1525))

  (CONNECT_DATA =

  (SERVICE_NAME = test)

  (FAILOVER_MODE = (TYPE = SESSION) (METHOD = Basic))

  ) )

  Or if you use failover and load balancing services you set failover described below:

  for 12 c: srvctl $ add orcl db-test - oel6vm1 favorite-available oel6vm2 - tafpolicy BASE - failovermethod SESSION service service - failoverretry 5 - failoverdelay 60

  for 11g: $ srvctl add service d orcl if test - r oel6vm1 - oel6vm2 BASIC EI SESSION m - z 5 AW 60

  Was what you mean?

• Hyperion 11.1.2.1 load balancing

  Hi all

  We intend to set up load with 11.1.2.1 balancing version.

  Server1 - Foundation, planning, Reporting

  Server2 - Foundation, planning, Reporting

  Server3 - HFM and FDM

  Server4 - HFM and FDM

  Server5 - database, Essbase (linux)

  Previously, we planning but load balancing has ever done foundation, hfm and fdm load balancing. So, need some suggestions before you start this process.

  For planning, we have installed planning (Server1 and Server2). In weblogic, we got 2 managed servers Planning0 and planning of the depenses1.

  By default the Round robin method, taken in load balancing.

  But the case of Foundation also, can we follow same process?

  can we use weblogic even serever1 only for all servers or can we install 2 servers weblogic 2 servers (server1, server2), where we plan to install Foundation?

  Thank you

  Mady

  You follow the same process you did for planning, weblogic will be installed by default on all servers, you can use the same domain server and weblogic administrator.

  Run the server configuration web foundation once completed.

  See you soon

  John

• FDMEE load balancing

  Hello

  According to the documentation of EMP 11.1.2.3

  Oracle Hyperion EPM system high availability and Disaster Recovery Guide

  I don't see anything about the FDMEE load balancing and high availability.

  Can implement us the high availability and FDMEE load balancing?

  Yes, it works in the same way as other Java Web applications.

  See you soon

  John

  http://John-Goodwin.blogspot.com/

• Content Tracker with load balancer

  We have 2 production servers and a load balancer.
  Each production server has its own database, stellent instance and components.
  So each prod server stores its own tracker data contained in its database.
  Question is what will be the effective way for the consolidated data of Tracker content of two servers?
  
  Experts, please help, thanks in advance...

  SQL Server or Oracle?

  You could replicate databases two product in a third, merged, only the database for single instance reports.

• NIC and load balancing

  Hello

  I have a new server ESXi 4.1. I currently have a cable connected to one of the two network adapters. When I put in place the management network, I selected the connected NETWORK card and left the other disconnect and deselect (via console settings directly on the physical machine). If I look at the network through vSphere Client cards, that's what I see: http://pix.sonhult.se/vmnics.png.

  What will happen if I connect another cable to my second NETWORK card? Will it set a new IP address for this NETWORK adapter? Could I have the load balancing or redundancy somehow?

  Thanks in advance!

  What will happen if I connect another cable to my second NETWORK card? Will it set a new IP address for this NETWORK adapter?

  Once you have plugged the network cable to your physical switch, you must assign to the vSwitch too. Once you do that the workload of the virtual machine and the management port will - by default - load-balanced in alternate way, which means that the virtual ports are distributed on the physical network interface cards.

  There is no need to assign an IP address any. An IP address is assigned only for the management port and the virtual machines within the guest OS.

  Could I have the load balancing or redundancy somehow?

  Yes. As mentioned above, this will happen automatically once you attach a second NETWORK adapter to the vSwitch.

  See http://www.vmware.com/files/pdf/virtual_networking_concepts.pdf for more details.

  André

• Network [vSphere 4.1] load balancing and failover vMotion

  GoodMorning everyone.

  I have some doubts about the configuration of the network for vMotion in vSphere 4.1

  I know that I can activate a single portgroup VMkernel for vMotion on the single host, but can I have this portgroup in a vSwitch with two or more physical NIC attached.

  Currently, I managed to have a vSwtich with vMotion and management of trade wth two NICs (see attachment).

  Both NICs works very well with the recovery for both exchanges.

  I would switch to a vSwitch with multiple network cards and I have some doubts.

  It seems that I could configure it... but the load balancer for vMotion supported on vSphere 4.1?

  And also, have the protgroup of vMotion in a switch with several network cards, is there a way to check what physical nic is currently used by vMotion?

  I am currently under a license from the company.

  Thank you all for your help.

  vMotion for vSphere pre 5 does not support Multi-NIC, in vSphere 5, you can configure several VMKernel enable vMotion and select different Nic Active and load balance between the two vmnic, but unfortunately not in vSphere 4.x

  For the management of ESXi, if you set active on both network adapters, you not to cross any active advantage / standby because it will only use on vmNIC, unless there is still a failure... but 1 vmNIC just because bandwidth is usually not a problem for the ESXi management only.

  So the answer: upgrade to vSphere 5 for multi-nic vMotion and don't worry ESXi management as an active NETWORK card is more than enough bandwidth,.

  Here's a YouTube on the Multi-NIC vMotion http://www.youtube.com/watch?v=7njBRF2N0Z8 configuration

• The order of failover and load balancing

  Hello

  I have the following scenario. An ESXi with 4 Gbps vmnic. The questions are:

  (1) if I have a group of ports configured for 'Route based on the original virtual Port code' in the policy of balancing load, and for the same port group I the option button 'Override switch failover command"checked, where I set up 3 of the active adapters vmnic, as well as the other vmic remaining as unused adapter, the ESXi uses the policy that I have configured (in this case 'Route based on the original port code') between the three vmnic load balancing marked as active? Or he uses them in the order that they appear in the section active cards?

  (2) Suppossed, I configured the four physical switch ports in an etherchannel group to use 'Route based on the IP hash' load balancing policy. In this situation, then I configured for a certain group of port to only used two active adapters and two others as unused? In this case, ESXi should balance the load using the method hash IP but only in two active adapters? Or it is a misconfigiuration and I should not configure my nic teaming in this way?

  (3) the official setup guide says "NOTICE on IP requires the physical switch be configured with etherchannel. For all other options, etherchannel must be disabled. ». How can I I configured my virtual network, if I have a few groups of political ports based on the hash of the IP to use load balancing and another uses 'Route based on the original port code. This is the case when I for example have two management ports using the same vSwitch with four vmnic (where they are configured as an Etherchannel in the physical switch). I would port one or several groups for virtual machines that use the IP of the hash method of balancing the load and vmkernel ports por management uses only a single adapter active with no back and as "based on the source port ID" load balancing as best practices said.

  Now, the four vmnic is the same for all traffic. The physical switch ports must be configured in an etherchannel group because certain groups of ports will use the method of IP hash, but others are not. The configuration guide I said SHOULD NOT use etherchannel if I won't use the hash IP method, but I'LL use it, but only in groups of one or more ports.

  Maybe I do not share the same vmnic from this situation.

  Finally, it's a philosophical question. What is the difference between 'The route based on the source port ID' and the 'road based on the source MAC Hash' load balancing policy? What is the purpose of the second? It is assumed that if I had two different MAC address in a virtual machine, it would be because I had two different virtual cards inside the virtual machine, which would be connected to two different port ID in the vSwitch, I can use the first strategy (based on the original port code). In other words, which would be the case where I had the traffic entering the same vSwitch but port ID with different source MAC address, so I should chose the method to distinguish the Source MAC address load balancing traffic?

  Thank you.

  Guido.

  (1) as long as you override vmnic only and don't change the policy for this group of ports, he uses the policy configured at level vSwitch and use the selected interface 3 with this policy

  (2) it should work, I don't think it's a problem for the switch receive packets on a subset of the aggregation. I do not think that Etherchannel is supported (IIRC, it is a Cisco proprietary protocol, VMware only supports LACP passive, which corresponds to the Port channel world Cisco.) Trouble me if I'm wrong!)

  (3) I think that's all right, as I have explained in 2), there is no special negotiations with the consolidation of VMware, the important thing only I know is to configure the port on the side of the switch channel if you decide to use the IP hash (that will lead to important questions)

  4) (self labeled) I think it may differ in some cases individuals, as when the operating system use the same MAC address for both NICs (aggregation in-vm) or if you advertise several MAC address for the same network card (ESX in a VM for example would make for its VM). Such cases differently affect this setting.

  That is the right question, and I'm curious to know if someone wants to develop on it!

• NIC Teaming: ID v-port-based load balancing

  Hey all,.

  According to VMWare ESXi does not support physical host interfaces LACP trunks.  I was told that the NIC teaming feature will allow you to specify which physical interface traffic is pushed through the source IP/VM database.  I was able to locate the NIC teaming settings in load balancing vSwitch properties, but I cannot determine how to set up a specific virtual machine, vNIC or source/destination IP address to use a specific physical NIC

  Can someone tell me how to proceed?  The setting of balancing says "Route based on originating virtual port ID"...  This isn't not always tell me how to assign a virtual interface to a specific physical interface.  Ideally, I would like to specify a destination IP address and a physical interface to use when accessing this IP address.  Simply being able to map a group of virtual machines to use a physical interface (without going through the VM groups on different vSwitches) would do well.

  Any suggestion is appreciated.

  Thank you!

  -Ben

  Intellectual property of hash based mode, 1Gbit/s physics 2 network cards can be effectively combined in 1 2 Gbps link?  Meaning regardless of VM, source/destinaltion IP/network, traffic etc will be shared between the two network cards until they are both is completely saturated?

  No, certainly not. It's like Weinstein explained. The NETWORK card used is based on the source and destination IP.

  You can take a look at VMware Virtual Networking Concepts that explains the different modes in detail.

  Route based on the hash of the IP ... Regularity of the distribution of traffic depends on the number of TCP/IP sessions for unique destinations. There is no advantage for the bulk transfer between a single pair of hosts.

  André

• Double connection ISP and load balancing

  Hi all

  I have the Cisco 2911 router k 9/s with 3 GB ports. I have also two different ISP connections, all have two different available bandwidth (one is asymmetrical, else a symmetrical).

  What I want to achieve is to ensure the balance of Nice load between two ISPS for all PCs behind the NAT device.

  What I know so far, it's that I can use CEF or PfR/REL. For both of this technology, I have some doubts.

  CEF: distributes the network load between the two connections based on sessions (which is good, because I strongly to use tools like Skype or Lync for audio/video conversations). However, what is happening, when I get on one of the ISP connections broadband bandwidth max? He's going to choke for 50% of the connections? Or it will detect the use of bandwidth and to force using second ISP?

  PfR/REL: as far as I understood it resolves my concern regarding the use of the connection, but what happens to the session? Should it also be based on this mechanism? As you know that it is very important for audio/video connections.

  Are there other tools that can provide these load balancing? I know DAB, but I don't want to decide manually, where each service (e.g. http or ssh) will have to go. I'm looking for something more automated.

  Thanks in advance for any help.

  Piotr

  Hello

  I assume that you have a static route for the subnet 213.192.65.0/24 on top of the output and with the combination of order

  network default IP 213.192.65.105 213.192.65.105 IP address is installed as a default gateway. What is the #2 ISP?

  If so, it explains why he always goes on ISP2 only.

  http://www.Cisco.com/en/us/Tech/tk365/technologies_tech_note09186a0080094374.shtml#flagging

  Just remove the config:

  Noneip default-network 213.192.65.105

  Noneip default-gateway 213.192.65.105

  Then again check the routing table:

  SH ip route

  Hope it helps.

  Best regards
  Akim

• Policy on NICs 1 Gb load balancing

  I use 6 x 1 GB NIC on the blades.  I have 2 NICs on board and 4 network cards provided by a single Mezzanine Card.  I have chosen a strategy of load balancing.  In view of the provision below, is there a reason to use load teaming instead of road based on the source port ID-based? I am allowed to use the load to team up if I want to but I don't know if there is no reason in this scenario-based:

  On Board 1 - Active MGMT, ensures vMotion

  2-edge - Production assets traffic

  Mez 1 - Active backup traffic

  MEZ 2 - Production assets traffic

  Martin 3 - Active backup traffic

  Mez 4 - Active vMotion, management standby

  If you have a license for LBT, use it. It is the only real policy load balancing that it monitors the load each NIC on a team and balances accordingly if there is an overload in one of the network adapters. Route based on origin Port ID just spreads the VM on the uplink (round robin) and does not care if one of the uplinks is most used and a used sub.

  I recommend using LBT on your Production and perhaps your backup traffic.

• NIC teaming with the IP hash load balancing

  Hi guys

  I have a virtual machine with a VMXNET3 10 Gbps. It is usually have a heavy traffic with a special server and saturate one of my two physical NIC of 1 Gbps in its PortGroup. I want to know what's going to happen if I turn on "NIC Teaming for this portgroup with basic IP hash load balancing". In my case the source IP address and destination are not change so the aggregates of traffic between my two physical network adapters?

  To avoid saturating completely the two 1GbE NIC, it can also be useful to look in charge based on the grouping and the NIOC. It will ensure that the other streams of traffic VM are not crushed by this machine a virtual when it saturates a network card. The disadvantage is that it requires Enterprise Plus licenses (using a dvSwitch).