Forum Discussion
Cory_50405
Aug 12, 2014Noctilucent
Typically end user traffic is directed to the virtual server, which then load balances the connection across the two backend servers. The return traffic from your servers will either route back out through the BIG-IP (specifying it as a default gateway for example) or the BIG-IP will use source NAT (SNAT) to ensure that all return traffic from the servers comes back through the BIG-IP.
Hope this answers your question. If not, keep 'em coming.
- gossett_151899Aug 12, 2014NimbostratusTypically end user traffic is directed to the virtual server. This is my man concept I can't get my head around. How does the end user traffic know where to go. For example i'm sitting on a server doing my work. How does it even know to check with the virtual server in the F5 system. Maybe i'm over complicating it I just don't understand. Does anything need to be set up on the servers that are in the load balancing pool?
- Cory_50405Aug 12, 2014NoctilucentTypically something like this: End user wants to access your application. Types the FQDN corresponding to your application into his browser, http://www.example.com for example, which DNS resolves to the IP address on your virtual server. BIG-IP then proxies the connection through to the backend servers. So DNS configurations would need to be in place as well, unless you are accessing the application directly by IP address. In this case, that IP address would be the virtual server address.
- gossett_151899Aug 12, 2014NimbostratusI think i'm still not explaining it right. What you said make perfect sense but this is what I need. No one will use the browser (that I know of; this may be where i'm confused). Basically I have two servers that do the same thing. Want to set up F5 so if one goes down it switches to the other one. How does the virtual server IP play a part in this kind of setup?
- Cory_50405Aug 12, 2014NoctilucentThat's the health monitoring and load balancing piece. So your virtual server will be assigned a pool, which will contain your two servers as members. You'll then apply a monitor to that pool which will be responsible for determining the health of the two servers. The monitor can be as simple as an ICMP ping, or it can be very robust to test actual service availability. If the monitor succeeds, then the server remains active in the pool. If the monitor fails, then the server is pulled out of the pool and requests are no longer sent to it. Requests will still be sent to the virtual server IP address, but the BIG-IP will just not proxy requests through to pool members that are not active (ones that are failing health checks).
- gossett_151899Aug 12, 2014NimbostratusAh Thank you!! So the health monitor will decide which server to be used based on its settings then ? And then like you said if one server/monitor check fails it will use the other server. Here's the second part of my question: if this is set up wrong will the servers in the pool be inaccessible? So lets say I set up the monitoring and load balancing wrong (since I've never done this). IF they are set up wrong will if affect the servers in question and make them unuseable ?
- Cory_50405Aug 12, 2014NoctilucentThe health monitor will determine which pool members are available to serve requests. The load balancing method (and any persistence configurations) you choose will be responsible for determining which pool member is directed the traffic. If you configure a monitor that does not accurately reflect the health of your servers, then yes you could have problems. For example, if your monitor is failing but the servers are actually available, or if your monitor is just doing a simple ping that is working, but the HTTP application on your web server is down. It's best to understand the application and build a monitor that would accurately reflect the health of it.
- gossett_151899Aug 12, 2014NimbostratusIf you configure a monitor that does not accurately reflect the health of your servers, then yes you could have problems would this actually cause the servers themselves to stop working though?
- Cory_50405Aug 12, 2014NoctilucentNope, it would just prevent the BIG-IP from routing user traffic to them. It could cause an unexpected DoS, depending on your setup. The servers themselves operate independently of the BIG-IP. It's only a proxy sitting in front of your servers. Keep in mind though if the BIG-IP is the only way that users access your servers, then a misconfiguration could make your application unavailable.
- gossett_151899Aug 19, 2014Nimbostratusso once I have it setup correctly how do I 'know' that the BIG-IP is load balancing ? I can see traffic but how does it work if one server would go down that the other would 'take over' so to speak. I have my monitor setup to watch http as that's all that's needed right now. How does the load balancing part actually work ? I'm sorry for such a basic question. I'm reading up on it while I ask as well so i'm sorry for my lack of knowledge ahead of time.
- Cory_50405Aug 19, 2014NoctilucentYou can look at pool statistics through the GUI to see active connections and such. You can also perform a tcpdump to see real-time traffic between the BIG-IP and your servers.