Forum Discussion
Kevin_Nail
Nimbostratus
Sep 07, 2007GTM and IP selection
Don't know if this issue has been addressed in this forum or not but...
We have an application that is fronted by LTM's in 2 different data centers. Both LTM's are reporting status to the GTM. In the event of a failure (failover) the GTM, of course, hands out 1 IP or the other. My question is, is there a way can we set it up so that the IP that the GTM hands out doesn't change?
We ave several external clients that cache the IP and then never asks for it again, so in the event of a failover the client application fails.
We would like the GTM to always hand out the same IP, regardless of which data center is active, so we can get around this issue. Like I said, I don't know if this is possible or not but any help is appreciated.
Thanks,
Kevin
16 Replies
- James_Thomson
Employee
The job of the GTM is the monitor the LTM and the LTM monitors the application instance. So, in your scenario, are you saying that the application fails, LTM notices, tells the GTM, then the GTM starts handing out the new available site, then the client fails because it never re-requests? If so, getting GTM to hand out the IP address of the broken application won't fix this will it? You application is down and the user's requests would fail anyway.
(If you want to do this, you don't need a wideip in GTM, you can just add a host A record) in Zonerunner since it is static, there's nothing dynamic about it. GTM will only hand out IP addresses that it sees as UP if they are defined in a wideip). The only static IP in a wideip would be the Fallback IP. You can select this as a fallback DNS response in the 3rd load balancing method)
In versions 9.2.X and later a feature called Route Health Injection was introduced. This allows the BIG-IP to participate in BGP, OSPF, or RIP to pull virtual servers in and out of dynamic routing so that IP address would move over to the second site and be hosted by your LTM's at site B. You'd need to run that same routing protocol on your routers and make sure to have the routing module on your BIG-IP LTM's. - Kevin_Nail
Nimbostratus
The job of the GTM is the monitor the LTM and the LTM monitors the application instance. So, in your scenario, are you saying that the application fails, LTM notices, tells the GTM, then the GTM starts handing out the new available site, then the client fails because it never re-requests?
Correct, this is exactly what is happening and I know that the GTM won't hand out the broken IP, that is not my intention. Ideally, I would like to have the IP be the same or move (as you suggested) to the other site, so that we do not have to depend on the client to re-request the IP. Just wasn't sure how to do that with the GTM. Thanks.- smp_86112
Cirrostratus
> Ideally, I would like to have the IP be the same or move (as you suggested) to the other site, so that we do not have to depend on the client to re-request the IP. Perhaps you are using the wrong product (GTM) for this. If you want intelligent failover behind an IP address that never changes, that seems like a job for the LTM, and not the GTM. With the LTM, you can use monitors and other features like priority groups to ensure connections are always routed to the available pool member, regardless of which data center it's in. This intelligence would all be masked behind a single Virtual Server address.
- James_Thomson_1Historic F5 AccountYou made a statement "We would like the GTM to always hand out the same IP, regardless of which data center is active". Wouldn't that be the definition of a static entry in DNS? If you add a static A record it would do that. I feel like you mean something else.
Solution 1 (not so great) The GTM can hand out two IP addresses at the same time, but there is no guarantee which one the client uses in which order. That is the problem with handing out two simultaneously.
Solution 2 (good) You can combine the GTM with BIG-IP Local Traffic Managers doing Route Health Injection to accomplish a scenario where the IP address moves (via BGP, RIP, OSPF) when all services behind that vip become unavailable. Route Health Injection means that if a service goes away, BIG-IP at site1 withdraws the route for that virtual and your routers in the network will push traffic to the secondary site still pointing at the same IP and the client does not need to re-request. This requires BIG-IP LTM at both sites with routing modules.- Meier_16843
Nimbostratus
What do you need GTM for in the second scenario?
- James_Thomson
Employee
You made a statement "We would like the GTM to always hand out the same IP, regardless of which data center is active". Wouldn't that be the definition of a static entry in DNS? If you add a static A record it would do that. I feel like you mean something else.
Solution 1 (not so great) The GTM can hand out two IP addresses at the same time, but there is no guarantee which one the client uses in which order. That is the problem with handing out two simultaneously.
Solution 2 (good) You can combine the GTM with BIG-IP Local Traffic Managers doing Route Health Injection to accomplish a scenario where the IP address moves (via BGP, RIP, OSPF) when all services behind that vip become unavailable. Route Health Injection means that if a service goes away, BIG-IP at site1 withdraws the route for that virtual and your routers in the network will push traffic to the secondary site still pointing at the same IP and the client does not need to re-request. This requires BIG-IP LTM at both sites with routing modules.- Meier_16843
Nimbostratus
What do you need GTM for in the second scenario?
- mganji_60652
Nimbostratus
Hi Experts,
May you please give me a document (or URL) regarding configuration of RHI in BIG-IP? or even a sample configuration? - dennypayne
Employee
LTM and GTM use ZebOS for the configuration of advanced routing protocols, Click here for the ZebOS configuration guide.
Denny - Hamish
Cirrocumulus
Does RHI require a separate license key? Or is it included in the base license? - johns
Employee
You will need a routing module license. - Hamish
Cirrocumulus
Ta. Thought so. - When using RHI + BGP to fail the same public vserver VIP between datacenters doesn't this imply that you are failing over at minimum a /24 of address space? I think this is the minimum size subnet most transit providers will allow you to advertise.
- Meier_16843
Nimbostratus
IIf you have multiple DCs you might also have your own AS inside which you run BGP
Help guide the future of your DevCentral Community!
What tools do you use to collaborate? (1min - anonymous)Recent Discussions
Related Content
DevCentral Quicklinks
* Getting Started on DevCentral
* Community Guidelines
* Community Terms of Use / EULA
* Community Ranking Explained
* Community Resources
* Contact the DevCentral Team
* Update MFA on account.f5.com
Discover DevCentral Connects