I believe that I understand the way that these work, however I don’t see any possibility for these rules to provide failover.
My setup: I have two ISP accounts with two static IPs (different networks); both accounts are connected to MT router. I would like to provide load balancing and failover capabilities in case one ISP fails.
Also what is the effect on apps like messenger, etc. I have read that they do not like load balancing.
I have used both of these scenarios, and I find that the first one load balances better than the second.
The plan that I am working through, is to use the netwatch tool to keep an eye on my gateways and when a gateway fails to run a certain script.
That script will then disable the load balancing routes and leave only one default route in place.
Trouble is im not too hot on scripts and its taking a while (any one got a script they want to share?) My other problem is that on one of my links which is a VSAT the gateway address is the satellite modem, and so that is on ethernet cable in the same room as router and rarely fails. My ISP is not being very helpful to find the next hop that I can use reliably for this netwatch tool.
I had the same kinda problem, but without balancing.
How I overcame your VSat gateway problem: Netwatch any site (google) and force the ICMP traffic to go through the VSat (using a rule). That way you do not need the second hop from your unhelpful friends. Just make sure that the google ping ONLY goes through the VSat.
From the wiki, I adapted the “improved netwatch 2” script for the failover.
am working on same kind of problem. Have two ADSL ISP’s. On this momemt only working to get the failover to work properly. In a later stage want to do the loadbalancing with failover backup as well.
The problem in this script: http://wiki.mikrotik.com/wiki/Improved_Netwatch_II is that whatever node I use to ping to (dns, ISP gateway, google, any other address) these nodes are never 100% up over time. When the ping doesn’t return the default route is beeing re-written and all traffic flows to my backup ADSL.
In itself not such a problem but my network is beeing used mainly for live communication like Skype, MSN, and streaming video. Everytime the route changes these streams are broken and needed to start up.
Sometimes it is only because the pinged nodes hickup for a while but in the mean time the ADSL in itself is not down… But the script rewrites the default gateway to go to the other interface to switch it back only after the next ping cicle.
My idea was to have the script pinging two or three nodes to decide if a route (and thus probably that whole connection) is down before the script starts changing the route costs.
but scripting is not my thing, I tried but don´t actually understand the give scripts I use. Anybody with any suggestions?
I’ll go through all these suggestions and see if I can’t come up with something that works. If anyone can provide a working failover script I (and apparently at least a few others) would really appreciate it.
I am working through the process, but like wirelessrudy scripting is not my thing either!
Can anybody post and example script or explain to me what the best way is to disable the extra NAT rules that are put in place for the Load Balancing system?
I have resolved the VSAT hot to ping, thanks to you guys, Used the tracert method and was able to confirm the vsat hub IP address. I thought this would be a better way than choosing a host like google.
If you still have no solution for this, you may use a hardware named “Multiple Homing Auto Balancer”. This hardware combines multiple gateways become one port to your Router, and it can balance all bandwidth from those gateways.
I never try this hardware yet, but a friend told me this hardware helps a lot. I am gonna buy one.
Hi, a good recommendation is to get an external load balancer. I would recommend TP-Link TL-R488T. It works great for my hotspot setup with 4WAN & 1 LAN.
You can buy dedicated LoadBalancers. They work fine (not always) but that’s to be expected, they are made for that task. I’ve used one myself.
But MT routers can handle the same stuff, it only needs some configuration. Browse this Forum and learn. This way you save yourself money and achieve some extra knowledge…
The advantage of using as much as possible MT stuff in your network is that you have plenty of different tools to monitor and work on it and it works everywhere the same. No need to find out how that $%& LB is working to see you can’t have real time traffice or real time loadbalancing going through. MT can do all!
Typically the satellite bonding boxes have TCP acceleration (TCP ack frame spoofing so that the sender will not wait for the real high latency ack frame to arrive before sending another TCP window). RouterOS does not have that, but we will look into that and see how hard it is to make.