Gateway Load balancing protocol ( GLBP)

now in this section we'll talk about gateway load balancing protocol it's one of the redundancy protocol which we can use apart from HS RP n vrrp now if we talk about the previous protocols we already discuss much more in detail about the previous protocols the major difference between the previous two protocols and this protocol is the GL BP support load balancing which means if you have multiple gateways which are connected in the lab and we can simultaneously forward the traffic through all the gateways but in case of previous tube methods which we used we can only forward one will be primary and all the remaining will be backup so which means at a time only one gateway can actually forward the traffic for one group now we can also we also did some advanced HSR implementations where we have created two groups one one primary gateway this will be primary for one group and this will be primary for other group but for a single group both cannot be forwarding but that is something what's supported in GL VP so we'll get into that more in detail in this section apart from that we also get into more in detail about this AVG and AVF concepts like in our HS RP we have two gateways like same and out of these two gateways one will be your active gateway which means which is going to actively forward the traffic and the second one we call it as standby in case of vrrp the terminology will be master and backup but here in case of GL VP we are going to call as AVG active virtual gateway and we have something called a VF if it's a VF stands for active virtual forum now the difference is a VG and AVF both are actively forwarding the traffic ellipse let's say let's say this is my physical gateway here now here AVG is the one who is going to control the Mac entries now when we say AVG so now here every router will be AV of active virtual forwarder means every router is going to forward the traffic every one every one is going to forward the traffic but there is one who is controlling all these active virtual forwarders or AV FCS AVG is the one who is going to assign the virtual MAC address to all the other ami ups now simultaneously we can have four gateways for one group and we just have only one what shall I pay for all these gateways let's say in my examples I'll be using 182 168 one or fifty but each and every gateway will be assigned as separate virtual Mac cutters so which means here if I have 4 physical gateways I will be having four separate virtual MAC addresses and different routers will have different virtual MAC address and this and send in answer to the ARP for the virtual like girls now AVG is the one who is going to reply to that ARP messages which are coming from the LAN and it's going to ensure that it goes from multiple gateways so it is like a controller who is going to control who is going to answer all your RP requests for that virtual IP and ensuring that it goes through your multiple AV offs so when it comes to configuration the configuration is more similar to your HRP configuration so you can see all the commands starts with GL VP and any group number you can use authentication command is more similar to that one preemption is by default disabled we can enable the pre and timers default is three and the day time is ten seconds but we can change it to one second and three seconds as per our requirement and the default priority value will be 100 but if you want a specific route or let's say I want to make sure that router 1 should be my AVG and router one router 3 should be my AVF and this will also will be a VF in that case I can I can change the priority value of my router to something more than 100 let's say 120 in my example and then you have to define the virtual IP now we see the configurations on the AVF there are three again the same configurations except priority value because we are not giving the priority value which means it's going to take the default priority value of 100 and 120 so which will ensure that your router one will be were active virtual acting as active or gateway at the same time it is also forwarding the traffic let's get into our lab here let's try to implement the similar configurations now for implementing the lab I am going to use the same routers which I used for all my previous labs so here were after one and the outer three will be acting as my virtual gateways which are connecting to my land on him zero by zero and I got one out of here out of four in fact I got multiple routers for verification I connected outer file as well as I connected outer 6 which is acting as mine and host and from this users we are going to verify to access some internet routes here and from router router 1 we have a default route router 3 we have a default route and from router 3 2 out of one and router now 2 2 2 1 n 2 2 3 we have a static route configure so the entire conversations goes similar what we did in over H SRP lab and then on this host I have a command called no IP routing because I am NOT going to use these routers as a normal routers they will be used as a host and then I have configured all these routers with a default gateway of 192 168 1.50 so there's something pre-configured in my lab here which which is the same like whatever I did in hitches RP so I'm not getting into that configurations so these things are already configured in my lab so I'll quickly verify here the connectivity between all the hosts first I will try to check from router one I'll try to ping to my other gateway which is 192 168 1 dot 200 that is route 2 3 I should see the reply so router 1 and 3 both the gateways or able to communicate with each other and also I will try to verify my host in the land which is 192 168 1.1 which is my router 1 here so I'm getting a reply and also I have one more host router 2 so you know trot out to its router 4 5 6 and 1.3 as I said I have multiple hosts connected in my LAN which is not mentioned in the diagram here so for verification I corrected multiple routers here this is my router fire and then I got router 6 so the IP addressing will be 192 168 1.1 191 6j 1.2 1 or 3 and all are conferred with a gate we have 192 168 1.50 as per my requirement so the first thing what we'll do is we'll try to do some basic GL VP configurations and it's very easy to configure just like go to the interface connecting in the LAN and all the command starts with GL VP group number now you can always use question mark to figure out what are the possible options so let's say I want to use authentication so I can say authentication MD fi any password Cisco so md5 question mark key string or keychain we also we can use some using key string as cisco is a password next thing these are optional commands not compulsory required then I can use IP to define the virtual IP I'll come to that IP load balancing also we'll see later on I'm going to enable the preempt to ensure that router 1 will come back to a VG it all it can become a BG if that goes down and come back again and I want to ensure that router one should be my my avg so i'm going to configure the priority value of 120 now the same command which i mentioned in the slide here so I'm just going with the same commands and then so i confident occasion preempt timers I'll also change the timers for faster convergence I'll say timers will be one is hollow and the date time is three seconds and then finally I'm going to configure the IP address and that IP is a virtual IP which I am going to use as one nine two one zero or not 50 done so once I configure this now automatically I should see some messages LBB messages you can see it's coming up now the same way if I verify to show history commands so I think similar commands I need to go and type on the router 3 except the priority value so I'll copy paste those commands here the preemption is enabled and then install priority well except that priority value everything I'm going to use the same commands on the router three now it's time for verification will verify first the GL BP configurations now for verification of GL BP I'll go to the outer one and I'll give GL BP you'll see a big output here in this output you can see the state is active the virtual IP address is one I do one Z should one not 50 and the hollow time was it's more like a similar output what we have seen in HSR P now Hector router standby routers here actor out his local standbys router 3 100 and the priority value of that router is this and the my local priority one is 120 it is also same and waiting will come to this waiting and load balancing options and here you can see there are two virtual MAC addresses created one is for these are actually group members group addresses here one is for one ninety one section 102 dot hundred so there are two forwarders we have forward a one he's our local router and the forwarder to is our router three and this is the MAC address of the forwarder to and this is the MAC address of the four one that outer one and after three and the preemption is also enable and minimum delayed is going to take 30 seconds to shift to alternate so there is a default timers so everything is same so even you can verify with a simple output shows LBP brief which is going to show you what is your virtual IP address and who is your local router and what is the virtual Mac and who is what is the remote router which is in listen state to dot hundred and the virtual MAC address off that autumn so for verification what I can do is I can go to my router host routers which is router for router of one out of six I will try to first verify my ARP table show IP ARP you can see already ARP table is this so I'll try to clear it now I'll try to access from router for so let's say let's try to verify by accessing YP ARP if I verify I can see my local MAC address which is one nine to one sixty one dot one on the router for I try to trace no one so directly I will trace it so that I can verify how the packet is traveling you can see first it goes to 100 which is my router one get VIN I'll go to my second host and I'll try to trace before i trace I'll just clear the ARP information because normally this ARP information has to be cleared in order to have a proper testing or verification so I'll say trace 12.00 1 and you can see it's going to so I need to clear it actual use so either you can clear Sorrells you need to shut down and no on the interfaces to get the proper output so I'll come to this load balancing method in general so the default load balancing it uses is round-robin so I will shut down the interface first and wait for the interface to go down and then no shutdown so in this way we can clear the ARP table properly to try to verify now I'll try to trace through and dot zero zero one I can see the packet going to 2 to 1 dot 200 last time it was 100 and then if I try on the router 6 again it will go why are Y again the primary gateway that is 100 so a round robin fashion is it's going to load balance I will come to that rod load balancing method so let us quickly verify what we did up to now up to now we just configure these commands where we have seen some of the basic BGP LBP implementations and then we are verifying the show DL VP output and finally for verification we can clear the cache and shutdown the interface notion on the interface that's the best way you can do the next thing what we'll do is we'll try to understand the tracking method like if you remember in in our HS RP and we are RP also with a tracking method let's take an example in case if this interface goes down let's say what I can do is if this interface goes down so now we'll try to verify the tracking configuration in ClpP now normally if you remember we have seen this interface goes down automatically water Pence's still that outer one is sending the hollow messages to router 3 so the router 3 will never come to know that this interface is going down there's the reason what I can do is I can track this interface so tracking is something what we did in our HS RP as well as we did in vrrp now similarly here also we can do but there is a slight change in the configuration so I can say it's there are a lot of changes in the configuration so if you want to track this interface the first thing you need to create a track object so I'm going to create a track object with a track object of 1 which is going to track that's s 1 by 0 interface and it's going to check the line protocol status so if that line protocol status changes down what I should do so the first thing we need to tree or create a track object with nmo one then what we are going to do is now here we are not going to change the priority value like if you remember in case of GL VP in case of previous two protocols HSR P and V RP if this priority value of the router one is 120 if the other outfits 10 i can automatically decrement 30 so that the priority value goes to 90 and then automatically it makes the router 3 as a primary gateway but here in digital VP it's not going to work based on the priority value it is going to work based on the weighted weight value weighting we can say and the default weighting of every router will be is 100 that is a default rating so what I can do is this is one way of configuration I'll go I will come to this configuration I can simply go to my interface I don't have this commands here so I'll write down I can simply say interface GL bp1 interface F 0 by 0 GL be p1 and I can say waiting whatever the waiting and then I can set track 1 and then I can give a command called decrement 100 so what I'm going to say is I'm going to say here is the default waiting is 100 and if that track object goes down that is what track object here so we are not getting into this configuration these different configurations something advanced so we'll come to this one now if this object goes down the track object goes down we are saying to decrement the priority value to 100 which means 100 minus 1 had become zero now in case of GL VP if any device is having the weight of zero which means it will not be a V of which means it will not be able to forward any traffic through that gateway this is one of the easy simplest possible solution which we can do if you want to ensure that if this interface goes down we don't want this router to forward the traffic so this is one possible solution we can do or or this is one solution I can say and the other solution is what we can do is we can use some configurations like I can configure the weight which is default is 100 and I can define the lower value let's say at the lower values for 90 and the maximum value upper value by default it will takes the same 100 and I'm going to say the decrement value of 10 which means it's going to subtract from 100 weight to 10 it becomes 90 now if any value if the weight value goes less than this lower value automatically that interface will not be able to forward and it ravish means it will not behave as a gate it will not forward any traffic that is one alternate way of doing the things so there are actually two different ways of doing the things and if you just want to make it very simple I recommend you to just go with a weight 1100 is a default and I can simply subtract 100 to make it 0 or else we can even define some lower values for that now it's all up to you how you configure so but these are the two possible ways we can configure the tracking on the LVP ok let us quickly verify this on our routers so I'll go with my second option or let me go with the first option which I discussed I will go to my router 1 and I'm going to track my router 1 saying that ok router 1 first I need to create a track object I'm going to subtract 1 I already have a track 1 not think so so if it is there I'll just remove track 1 and then I'm going to say interface track the interface I want to track my silly interface which is connecting to my service for her and the line protocol status done so the track object is created and then now I need to link this track object with my GL BP so I'm going to say al bebé now weighted track wait a maximum value so we can define the weight value maximum what you want by default is 100 if you don't define it so I'm just going with 100 and I'll say track I'm going to define the object and I can define the decrement value if you don't define the decrement value there is one more thing in case let's say in my previous configurations here in this example I did not define the decrement Val let's say if I don't define the decrement value the default decrement value will be 10 the default decrement value will be 10 but if you want you can define any other values so here I'm not going to use then I'm going to use a teef decrement 100 totally so I'll go to router 1 and I'll say decrement value it will be 100 so the default weight is 100 I am NOT applying that if you want you can even apply that so it's not mandatory you can simply say 100 but I'm not giving this command I want to show you that so the default rating is 100 and then decrementing also 100 if that interface status goes down okay so verification so if I could show GL VP you can see some configuration here the waiting is 100 which is default which I just discussed the default value will be 1 and upper value value 100 the default values and we right now we are tracking the object 1 and the state is up in case if it is down it is going to decrement automatically 100 so let's make it down ok so what I'll do is I will go into s 1 by 0 interface which we are tracking I am going to simply say shut down so when that interface goes down automatically now still the outer one is a VG remember on a router one will still will be active gateway which is controlling your ARP request because still it is sending the hallo messages on router wonder router 3 but it is only not forwarding the traffic so if I can show GL BP once again for verification you can see weight is zero which is very low right now the reason is track object is done and decrementing the value now for verification you can try going to router 3 is 4 5 6 and you can try tracing probably all the trace packets will go from router 3 to reach to a lot network so I'm not doing that because sometimes to have a proper exact exact output it's you need to you need to really shut down the interface and you have to clear that they are requests it takes time so I'm not going to show you that but if you want to verify try shutting down the interfaces of 4 5 6 and bad to move shut down which is going to ensure that your ARP will be cleared and if you try to trace from 4 5 6 all your trace package will go from router 3 and outer one will not at all for what any traffic and the reason is the default wait is the D weight is 0 so it's not eligible to do that so the other way which I discussed I will quickly show you that other method as well sure an interface F 0 by 0 I'll try to modify the configurations on this interface and I'll say and I will remove this tracking and I will reconfigure it again so I'll say LBP 1 wait I'm going to say default weight is 100 and I am going to define the lower value if it goes less than 95 that's same means automatically it is not eligible to forward the traffic less than 95 the default is 0 I am changing to 95 and then I'm going to say LBP 1 waiting track 1 and then sorry no track on your track 1 and at this time I'm not defining any decrement value because if you don't define a decrement value default it will decrement by 10 so I'm just leaving it blank so I'm using the default values for your understanding but it's all up to you - you can use any way you and whatever the values you use hue so we need to ensure that if it subtracts 10 automatically becomes 90 which is less than mile less threshold value which which makes your outer one not for with the traffic some not defining any values so if I say show GL BP now straight is down because the interface is still down I did not make that s 1 by 0 it's still in down straight now it is decrementing the 10 value which I said default the lower value is 95 and now the weight is 90 which is low then what is lower value define so in this case also it is not going to format the traffic so there are two different methods actually so it will be little bit confusing but there are two different methods you can you can actually configure the tracking in GL baby so I want to ensure that you should know both these methods the next thing we'll see is load balancing ok so let's talk about load balancing in GL people like I discussed that GL VP is one of the protocol which is only supporting load balancing the only protocol which support load balancing let us see how it is going to do that so I'm going to take four physical gateways which are acting as one virtual IP address and we have for virtual MAC addresses and assume that there is one router which is a VG and all the remaining routers are ABF's now what is my requirement is the default load balancing method will be round robin round robin is a default method so round robin with each macorís is used to respond to arp request in turn which means everyone will have an equal turn to actually answer the IP request and everyone will have an equal chance to forward the traffic that is what round robin method whereas apart from the round robin we can even define something called host dependent load balancing where in case of host dependant load balancing let's say if this user is trying to form the traffic from this gateway and every time the traffic coming from this arp this this user it always uses physical gateway so a given host will be always uses the same router I can also do that but if you have less number of hosts generally if you have more number of hosts we don't go with this option so it's all up to you how you want to use but there is one more method called weighted load balancing in case of weighted load balancing we can define some weight parameters which is going to do load balancing based on the applied weight let's say let's take an example I'm going to say the weight of this router is 100 or I can make it as 200 here and then the weight of this is 100 and the weight of this is 50 and the weight of this is 25 so now if you if you have high-speed link let's say this this link is very high speed link now if you compare the weights here now this ear router let's give some names ABCD now a gateway is going to forward two times more traffic than B because because of the weight is double man that and similarly let's say if I take one one packet it is going to send one ARP message or one host from this side it will be sending two and it will be sending four and this will be sending eight so if you see which means it is going to send eight times more than what Dee is actually sending so that is something what we can do weight based now the default weight as I discussed all the routers will have the default weight of 100 but what I can do is I can change the weight as per the bandwidth requirement let's say if this link is having very huge amount of bandwidth I can simply increase the weight something 200 to this router so that it forwards more traffic rather than comparing to other gateways now to configure this the command is very simple here we just need to go to interface F 0 by 0 and we have to say it's your BP load balancing when I when I use Ivor's help we can find all the three methods here now the default method will be this for verification you can go to my command line I can go to my router 1 and I can simply say I'll make the interface up which is shut down actually in the previous for example we did that so I'll say show gel BP I did not change any method if you you can verify my interface configurations I don't have any load balancing command here which means the default load balancing method will be round-robin but if I want I can change my load balancing method to the command will be a load balancing and then I can simply say weighted so weighted based on the weight it is going to do or I can simply say host dependant as per your requirement so it's all up to our requirement we can use any one of this load balancing method so so finally what we discussed about the gel VP we have something extra information regarding the gel VPS the main thing is it is going to forward the traffic multiple gateways will be used to form the traffic and it is going to have only my virtual MAC addresses like you can have up to four physical gateways in one group and they will be having only one virtual IP and all will be forwarding the traffic so there is a one advantage we get in zeal VP you

Channel: Sikandar Shaik

Views: 42,216

Rating: 4.8888888 out of 5

Keywords: GLBP, Gateway Load Balancing Protocol, redundancy protocols

Id: 4jD75kemvW4

Length: 27min 44sec (1664 seconds)

Published: Sat Apr 19 2014