Discussion on load-balance and load-sharing

Hi, I found a article, which discuss the difference between load-balance and load-sharing. I think the explanation is pretty good, please see below. But I still have a question: how can we decide to choose one the both balance in the production environment ?  Thank you
"In short, load balancing tries to distribute traffic evenly over multiple paths, whereas, load sharing intends to do it (for the lack of a better term) equally.  True load balancing is difficult to achieve.  For example, let's say there were two links (100 mbps and 300 mpbs) and a router needed to send out 600 mbps of traffic.  Load balancing would distribute the traffic evenly, sending 300 mbps on each link.  On the contrary, load sharing would divide the traffic equally based on the available resources, sending 200 mbps on the slower link and 400 mbps on the faster one. "

Disclaimer
The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.
Liability Disclaimer
In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.
Posting
That's not how Cisco uses the terms, and generically they are often used almost interchangeably.
Cisco uses load balancing as the catch all for how a single L3 device routes across multiple paths to the same destination.  Equal metrics or equal actual load distribution are not required.  Most often, load balancing will be discussed with ECMP, but unequal path loading balancing will include Cisco's proprietary IGPs, such as EIGRP.
Cisco uses load sharing when using multiple paths when a single L3 devices doesn't normally route across multiple paths or multiple L3 devices are involved.  Cisco load sharing discussions usually revolve around BGP.
Generically, I would say load balancing has more of a dynamic aspect to it, i.e. something is trying to actively balance traffic across multiple paths, while load sharing might mean multiple paths are utilized but not actively dynamically balanced.
I'm unsure what's your question with a production environment.

Similar Messages

  • Load balancing and session sharing

    Hi,
    in my company we have the following situation:
    We have 2 instances of the same web application on differend servers.
    An external load balancer (Alteon) routes the requests to these instances.
    The problem:
    If one instance is shutting down, all the sessions of this app are gone. These users has to login again.
    The question:
    Is it possible to avoid this?
    Is is possible to keep these session, so the users don't have to login again?
    Maybe with a kind of session sharing between these 2 instances?
    using:
    oc4j standalone - 10.1.3.1
    OS: Windows
    thank you
    Marcus

    The solution:
    Multicast IP replication is the default replication protocol used in a standalone OC4J installation
    see: http://download.oracle.com/docs/cd/B32110_01/web.1013/b28950/cluster.htm#BABBHGHJ
    -> Configuring Multicast Replication

  • Load-balancing vs Load-sharing (L2/L3)

    What is the difference in load balancing and load sharing. Can you generalize load-balancing to layer two? Or layer three protocols also do this, except BGP?
    FHRPs like VRRP, GLBP, and HSRP are essentially load sharing protocols (except GLBP though). Loop prevention mechanisms like MSTP, G.8032, and REP also do not do load balancing I guess. Instead they offer a way to distribute (share) traffic over multiple links unevenly. Why do people still call it load-balancing?
    While link aggregation may truly be considered load balancing. Where traffic can be balanced based on a few attributes ( src-dst/mac, ip, port).
    And what role does fast-switching/process-switching play in load balancing. I am aware that these are the very processes that ultimately do load-balancing, given enough equal cost paths. Does CEF really take load balancing to another level? It still does the same job, but by caching flow data (RIB) to the CEF table, doesn't it?
    Incase of IGP, load-balancing may be possible, but BGP just doesn't support it. Because it only selects one best route (?). Even with multipath, BGP advertises best path only; it does install multiple routes in routing table though. So IGP does the balancing?

    What is the difference in load balancing and load sharing. Can you generalize load-balancing to layer two? Or layer three protocols also do this, except BGP?
    FHRPs like VRRP, GLBP, and HSRP are essentially load sharing protocols (except GLBP though). Loop prevention mechanisms like MSTP, G.8032, and REP also do not do load balancing I guess. Instead they offer a way to distribute (share) traffic over multiple links unevenly. Why do people still call it load-balancing?
    While link aggregation may truly be considered load balancing. Where traffic can be balanced based on a few attributes ( src-dst/mac, ip, port).
    And what role does fast-switching/process-switching play in load balancing. I am aware that these are the very processes that ultimately do load-balancing, given enough equal cost paths. Does CEF really take load balancing to another level? It still does the same job, but by caching flow data (RIB) to the CEF table, doesn't it?
    Incase of IGP, load-balancing may be possible, but BGP just doesn't support it. Because it only selects one best route (?). Even with multipath, BGP advertises best path only; it does install multiple routes in routing table though. So IGP does the balancing?

  • Network Load Balancing and failover for AFP Sharing

    Dear all,
    Somebody kindly teach me to use round robin DNS to perform the network load balancing, it's success but not the failover.
    I have 4 xserve and want to do the load balancing and failover at the same time.
    I have read the IP failover document and setup it successfully, but anyone know is it possible to do the IP failover for more than 2 server?
    For example, 4 server serving the AFP service at the same time, maybe I have 1 more extra server to do the IP failover for thoese 4 servers.
    As I know, IP failover require Firewire as the heartbeat detection. But one xserve only have 2 firewire ports. May I setting up the IP failover only by a ethernet port and an IP address? does it possible to detect and failover to any server after server down has been detected?
    I believe load balancer maybe the best solution but its cost is too high.
    Thanks any advance!
    Karllee

    well, u have 2 options here
    software load balancing
    request comes it foo.com -> ws7u2 hosting foo.com is configured to run as reverse proxy . this server sends any incoming requests to one of the four back end web server 7 handling your incoming request
    hardware load balancing (this you need to invest)
    request comes to hardware load balancer who responds for foo.com -> sends requests to four ws7 server hosting your application
    you could try out how software load balancing works out for you before you invest in hardware load balancing
    here is more instruction on configuring ws7 + reverse proxy (software load configuration)
    - install ws7 on foo.com
    - create a new configuration (choose port 80, disable java

  • Network Load Balancing for AFP Sharing

    Dear all,
    Anyone can kindly teach me how can I configure network load balancing with 2 xserve?
    Currently I success to bond 6 ethernet port with a virtual IP in a single machine and I have a link aggregation setup in my switch. It works fine.
    How to configure 2 xserve with 6 ethernet port per each to have a single virtual IP?
    my switch do not support link aggregation with a virtual IP to do the load balancing so I just can consider to do it in software level.
    Anyone know whether OS Leopard/Snow Leopard can do it? Or any suggestion for 3rd party software can perform this?
    Thanks expert!
    Karl

    It sounds like you need a load balancer. There's nothing built-in to Mac OS X Server that's going to support one virtual IP address shared across multiple physical servers.
    Your problem, though, is likely to be one of throughput - I don't know any cheap load balancers that will sustain that kind of throughput, so you may be looking at some serious $$$$s.
    There are some software-based load balancers that might be able to handle the load balancing side of things but many of them are designed around HTTP so might not work so well for other protocols.
    In addition, the software load balancer is going to suffer the same bottleneck as your AFP server, but doubly so - two servers with 6 x 1gbps links each you have a theoretical limit of 12gbps.
    To run that through a load balancer, the load balancer will need double that - 12gbps for the client-side, plus 12gbps for the server side. In reality you're probably looking at needing 10gbps interfaces and switches if you're really pulling that much bandwidth.

  • Cisco Load balancer and Web Dispatcher to the same portal

    Hello Experts,
    We have implemented intranet portal with Cisco as the load balancer. Now we need to expose this intranet to the outside world as an extranet portal. So the same portal will be accessed from both intranet and from outside. We are thinking of installing a web dispatcher in the DMZ so that outside users can access the Web Dispatcher URL to access the intranet portal. In effect intranet users will use load balancer and extranet users will use Web Dispatcher to access the same portal. Now my question is if we configure Load Balancer and Web Dispatcher to the same portal, will the portal be able to load balance properly? Is this the right approach?
    Thank You,
    mansooralip1

    Dear Andrew,
    We need to provide access to our intranet to some outside companies for them to also use some of our portal applications. As per your answer, I understand that I can configure Web Disptacher to talk to the Cisco Load Balancer of our portal. In this case Web Dispatcher will work just as a reverse proxy. But when I discussed this with one of our basis resource, he told me that when we install and configure Web Dispatcher, it always ask for the Message Server URL and Port number, even if I just want to use Web Dispatcher as a Reverse Proxy. If his concerns are valid, I do not think I will be able to configure Web Dispatcher to access the cisco Load Balancer because I cannot put Cisco load banacer URL and port instead of the Message Server URL and Post Number. Can you kindly share your comment on the same?
    Now the second part of my question, if Web Dispatcher cannot be configured to talk to Load Balancer(as mentioned by our basis resource), I will have to use two load balancers. One web Dispatcher in DMZ as a Load Balancer *** Reverse Proxy for the external users. Second the internal Cisco Load Balancer for the intranet users. So the same portal will be accessed by two load balancers. My question here is, in this set up, can the portal work efficieintly here by distributing equal loads two both the server instances?
    Thank You,
    mansooralip1

  • Load balance and Back up

    Hi,
    We have configured load balancing for a serial link in OSPF. We are using one PRI channel for that. Is is possible to configure a back up for this serial link from PRI channel. ( using unused channel from PRI ).
    we have used follwoing commad in serial link configuration.
    backup load 60 40 ( for load balancing )
    backup interface dialer 100 ( one of the PRI channel for load balancing ).
    can i asign one channel from PRI as a backup if this serial line fails. Or if serial line fails load balance channel will trigger and act as a backup ? Can i configure both load balance as well as backup , if leased serial line fails.
    Thanks in advance
    Bapat

    Bapat
    When you configure load balancing (or load sharing) you automatically get backup. With load balancing there will be two separate paths to the destination in the routing table. If one of the paths fails and is withdrawn from the routing table then all traffic will be sent on the remaining path.
    The configuration that you have posted here is defining a channel of the PRI as backup (with the backup interface command) and by using the backup load command you have also indicated that if the primary link gets over a configured load threshold that the PRI should activate and share the load.
    So this gives you both backup and load share.
    HTH
    Rick

  • VPN load balancing and ASA !!!

    Hi netpros,
    I have a couple of questions about this and hope you might be able to assist me.
    1.- Are VPN load balancing and failover (Active/Active) mutually exclusive ..? I mean they can't be used at the same time correct ..?
    2.- How does the ASA handle the return traffic from the Internal LAN towards the remote client .. Because the cluster only requires ONE public virtual IP address, which will work for incoming packets .. but what about the return traffic which has knowledge of the DHCP scope's default gateway IP address only .. ? How gets the returned packet redirected from the default gateway IP address to the respective ASA internal IP address .?
    3.- VPN load balancing only applies to remote clients using easy VPN technology (easy vpn client, hardware client , pIX using easy vpn client etc ) and does not work with static LAN-LAN tunnel .. correct ..?
    Your comments are much appreciated

    Hi Gilbert ..
    1.- Thanks I wanted to make sure.
    2.- I know that .. my question is in regards the return packets .. for example if I have the below IP schema:
    ASA1: Public 20.20.20.20
    Private 192.168.1.1
    ASA2: Public 20.20.20.21
    Private 192.168.1.2
    Cluster virutal IP: 20.20.20.10
    Default gateway for segment 192.168.1.0 is 192.168.1.1
    Let's say that a vpn client tries to connect and the cluster instructs the client to connect to ASA2 20.20.20.21. The packets reach the internal server at 192.168.1.100. The internal server then sends the return packets back to the client by forwarding them to its default gateway which is 192.168.1.1 (ASA1). Here is my question .. how does the cluster handles this because the return packet are supposed to be directed to ASA2 192.168.1.2
    3.- Any idea about this one ..?
    Cheers,

  • Load balancing and rfc metadata repository in reciever rfc communication ch

    hi.
    i want to know the purpose of load balancing and rfc meta data repository in RFC communication channel.
    and can u send me any examples on this load balancing.
    waiting for your response.
    bye.
    regards.
    seeta ram.

    Hi Seeta Ram,
    Load distribution is handled by the message server (there is one message server in an SAP System). When a user logs on, the message server assigns him or her to the application server that currently has the <b>smallest load</b>.
    Well now you can understand that we use load balancing for better performance by distributing the work to different processes to balance or maintain the work load in SAP system.
    For more information refer to this link
    http://help.sap.com/saphelp_nw04/helpdata/en/28/75153a1a5b4c2de10000000a114084/content.htm
    Regards
    Sumit Bhutani

  • Advantages of using a webserver inbetween a load balancer and application servers

    I am building out a new weblogic domain.
    I am wondering which one of these configuration to go with:
    1. Load balancer > weblogic servers
    2. Load balancer > web server > weblogic servers
    Could someone tell me what are the specific advantages of having web servers inbetween a load balancer and application servers (besides caching static data content and acting as a proxy)?
    Thanks in advance
    Srini

    Other than hosting the static content, nothing much really.   We have our load balancer go straight to WL for applications without static content and route to web server if there is static content.   Easy enough to do it both ways, best of both worlds.

  • For a true load balancing and high-availability OHS, OPMN, and mod_oc4j

    i have read this link of Enabling Clustering on oc4j9.0.4 standalone app server
    http://www.oracle.com/technology/docs/tech/java/oc4j/htdocs/getstart.htm#1015479
    To test the clustering, start up the load balancer by executing "java -jar loadbalancer.jar".
    C:\OC4J_EXTENDED\j2ee\home>java -jar loadbalancer.jar
    In a future release of Oracle Application Server, loadbalancer.jar will be
    desupported. Because of this, we strongly suggest that you discontinue your use
    of loadbalancer.jar in this release. Under high loads, loadbalancer.jar may not
    function properly. For a true load balancing and high-availability solution,
    please move to use OHS, OPMN, and mod_OC4J. For more information, please see
    http://otn.oracle.com/products/ias/ohs/content.html
    Balancer initialized...
    what load balancer should i use for web clustering
    <frontend host="balancer-host" port="balancer-port" />
    balancer-host=localhost
    balancer-port=80
    for all nodes i mentioned same host and port in http-web-site.xml.Is it correct?
    i completed all the steps and run http://localhost:6666/session/SessionServlet
    i hit 3 times
    in the different browser http://localhost:7777/session/SessionServlet
    instead of coming 4 it starting from 1 only.

    can i use this loadbalancer.jar or not?
    how to mod_oc4j in standalone app server

  • Load balancing and High Availability topology

    Our Forms 6i client-server application currently runs on Citrix farm of 20 Windows 2000 boxes (IBM Blade Servers 2 CPU and 2 Gig Memory).
    Application supports 2000 users.
    We are moving to AS 10g r2, forms 10g and the goal is to use same hardware, 20 Windows boxes (or less), for intranet web deployment.
    What will be our best choices for application Load balancing and High Availability?
    Hardware load balancer, Web Cache, mod-oc4j? Combinations?
    Any suggestions, best practices, your experience?

    Gerd, I understand, that you are running 10g web forms through the browser, but using Citrix for deployment. This means that in addition to Application Server and Forms runtime sessions, it will be separate browser session opened for each user. What the advantage of this configuration?
    Michael, we are aware, that Citrix is not supported by Oracle as a deployment platform. That only means that prior contacting Oracle Support we have to reproduce the problem in standard environment. It was never been a problem to reproduce problem :) We were using Citrix as a deployment platform for Forms 6i client/server for 4 years, but now we are forced to upgrade to 10g.
    We are familiar with various Load balancing options available. The question is which option is the most "workable" in our case.

  • 2 ISP load balancing and redundancy

    Hello!!
    Our small company has about 40 branches spreaded within city. Branches are connected by optic wire supplied by our ISP. So in ISP our branches are located in one VLAN. From every branch we created VPN tunnel to our server room in central office. Central office is like a cetner point. If optic wire fails to central office, there would no VPN tunnels and no network to all branches. Moreover, all the traffice goes through central office.
    Now we decided to pave one more optic line to our central office. And that will increase bandwidth and redundancy.
    Private network topology: There are no default gateways and ip-addresses. For examle, at first branch I will plug computer directly into media converter and at the second branch plug another computer to the media converter. After that this two computers became in one network. And can assign any ip addresses to them.
    What I have: our firewall do enough work, don't want to overload it. But we have some free ports in our new cisco 3750. The question is how to do load balancing and redundanccy? Can it do load balancing according to traffic? And how load balance incoming traffic? For example, connection was established from branche's router, how this router will choose through which line make connection? By the way, at all branches we use noisy cisco
    3700 series routers.

    Sorry for upping 1 year old threat.
    We talked to our Network Provider. They said "these two cables are coming from two different places, so there is no way to use etherchannel. You must use active-standby solution."
    Relying on STP we just put two cables into 3750 stack. But with default STP settings, connection was very unstable, many packet losses and disconnections. So we found easy solution with "flex links", making one interface backup of the other. And only now I recognized that this is not a failover solution. Because, if network beyond media converter will down, link from media converter to switch would still up.
    What could I do to make our L2 WAN redundant? Are there any additional STP settings.

  • Load balancing and Failover

    Hello,
    We are wondering how load-balancing and failover of tpcall() work with
    WTC:
    The scenario:
    We have one WLS Domain and two Tuxedo Domains. The Tuxedo Domains offer
    the same set of services.
    In the bdmconfig.xml, we specify connection_policy as 'ON_STARTUP' for
    both Remote Tuxedo Domains. We also Import (T_DM_IMPORT) the same
    Tuxedo Service from both Tuxedo Domains.
    Questions:
    1. Is there any load-balancing of the tpcall between the two Domains? If
    so, is it round-robin? If round-robin, what determines the order?
    2. If it is ONLY Failover, what determines the order of the tpcall? And,
    is the Failover automatic? Or do we need to code for retry on failure?
    3. ON_DEMAND vs ON_STARTUP: Does ON_DEMAND drop the connection to the
    remote domain upon tpterm? And does ON_STARTUP use a pool of
    TuxedoConnection objects?
    4. Are there any configuration parameters for
    'max_number-of_connections? What determines how many simultaneous
    connections can be made?
    Thanks,
    Suresh Mohan.

    Hi Suresh,
    The following are my answers to your questions.
    Suresh Mohan wrote:
    Hello,
    We are wondering how load-balancing and failover of tpcall() work with
    WTC:
    The scenario:
    We have one WLS Domain and two Tuxedo Domains. The Tuxedo Domains offer
    the same set of services.
    In the bdmconfig.xml, we specify connection_policy as 'ON_STARTUP' for
    both Remote Tuxedo Domains. We also Import (T_DM_IMPORT) the same
    Tuxedo Service from both Tuxedo Domains.
    Questions:
    1. Is there any load-balancing of the tpcall between the two Domains? If
    so, is it round-robin? If round-robin, what determines the order?Yes there is a load balancing between two remote Tuxedo TDomain Gateways.
    The algorithm is random, not RR. Over time this should give equal
    opportunities to both remote TDomain.
    >
    2. If it is ONLY Failover, what determines the order of the tpcall? And,
    is the Failover automatic? Or do we need to code for retry on failure?The load balancing is always there. The failover is automatic. When a
    connection to a remote TDomain encountered a problem (ie network) the remote
    domain will be put on retry open connection (in ON_STARTUP) and the load
    balancing will not select it until the connection re-established.
    However, the tpcall() that encountered the error will not be retried to send
    to different destination. It is up to the application to decide whether it
    want to resend. Any requests called after the error will not select the
    failed Remote TDomain.
    >
    3. ON_DEMAND vs ON_STARTUP: Does ON_DEMAND drop the connection to the
    remote domain upon tpterm? And does ON_STARTUP use a pool of
    TuxedoConnection objects?TPTERM() only terminate your application session to WTC. WTC still maintain
    a secured T-session to remote Tuxedo TDomain. WTC does not use a pool of
    TuxedoConnection Objects, the object stored in the JNDI refers to WTC.
    >
    4. Are there any configuration parameters for
    'max_number-of_connections? What determines how many simultaneous
    connections can be made?No. As described in #3, there is no need to use connection pool in WTC. WTC
    uses session and virtual circuit design concept as Tuxedo TDOMAIN, the
    logical pool is created/destroyed dynamically. That is the reason why you
    can have a lot of TPACALL() outstanding at the same time. (The limitation is
    the availability system resource.)
    >
    >
    Thanks,
    Suresh Mohan.Regards,
    Hong-Hsi :-)

  • Load Balancing and Failover with 10G Standard Edition

    Hi,
    I am new to Oracle Replication and need some help setting up replication for load balancing and failover. Is this possible using Oracle 10G Standard Edition? I plan on having all updates done on the master site and both databases will be for reads. In case of failure of the master site, I would need to be able to failover to the other database.
    Also, if anyone knows of any documention for Basic Replication in 10G, please let me know.
    Thanks.

    Simple nnapshot replication of data would require significant manual effort to configure to load balance or failover. One the load balancing side, you would generally be limited to to static load balancing-- assigning half the users to one machine and the other half of the users to the other machine, regardless of who is actively using the machine. Failover would be a significant manual effort, particularly to bring the failed machine back into the cluster. You would be implementing the guts of multi-master replication.
    Frankly, if you actually have a system which is valuable enough to need load balancing and disaster recovery, I'm going to wager that it will be far cheaper even in the short run to buy more boxes and/or enterprise edition licenses than to try to implement this sort of thing yourself. In the long run, it will be far cheaper, since it will be far easier to maintain. Building all this yourself would probably be penny wise and pound foolish.
    Justin
    Distributed Database Consulting, Inc.
    http://www.ddbcinc.com/askDDBC

Maybe you are looking for

  • My vocal track has sped up for no reason

    I was mixing a song today, went to the beginning of the song to listen to it and when the vocals came in they were about twice as fast as they were recorded and had previously played back. i had probably listened to it 10 times and it was fine, then

  • Wrong port in FTP receiver adapter - no errors in message monitoring?

    Hello all We're on XI 3.0 SP16 and have created / configured a HTTP XML -> XI -> FTP Server scenario. In the receiver comm. channel (ftp) we had specified a wrong port number and always wondered, why we never got errors in message monitoring. Everyth

  • Phone locked up- and I am ready to throw it out the window

    Last night before going to bed, I plugged my phone in to charge. (Using the IPhone charger and plugged into the wall). My phone is currently showing the battery, yesterday's date and the time of 9:22pm. I can't get the phone to do anything. I just he

  • Purchase order creation with reference to Contract

    Dear All                 Purchase order creation with reference to quantity contract the contract quantity is not copied in to po. any configuration copying rules required. please guide Regards Anand.

  • Help in the query

    hi folks, Help me to build the query here. Tables BKPF - fields belnr, blart        BSEG - fields kunnr,wrbtr, zuonr I need to get the data into a single internal table whose strucutre consists of fields kunnr, blart,wrbtr and zuonr the common field