Forum Discussion
NimbostratusGTM Setup question with HA design
Wanted to reach out for a design and setup question.
We have normally setup standalone GTMs in each datacenter.
We normally have something like this: dc1dc2 GTM GTM LTM LTM LTM LTM
With the new DC design, we have a pair of devices that have GTM/LTM. They have collapsed the equipment.
There are just two devices and are in a HA Active/standby pair.
Does the GTM listener point to the floating ip or is it just a VIP on that F5 virtual server network point to the self ips of the devices?
Is there a document you could point me to?
Would DNS be listening on Two ips or just one?
Thx
2 Replies
Dave_McCauley_3Cirrostratus
When you create the listener, you'll see it's just a VIP that gets created. That VIP will be in the traffic group that fails over between the two devices. You could also have it use the destination IP of the floating IP if you're short on IPs but typically you'd give it it's own IP.
You could also create the listeners on the local traffic group for each device and have LTM VIPs failover between the two and have the GTM listeners always be active.
Honestly if you could do vCMP and separate them out, or have separate devices you'll be happier in the long run. Combining GTM and LTM gets messy in my experience, as you're finding out trying to architect this.
--D
this depends on your business need.
1. If you want GTM listener be sort of active-standby then you create any listener IP and this IP will become part of traffic-group-1. DNS traffic in that case will be sent to currently active box. However, from GTM perspective of view you still have two GTMs, since GTM can not be active or standby by nature. It can be primary or secondary but always active. If you use build-in BIND, such type of config will not work for you, since BIND will listen on non-floating self-ip only.2) You can crate two gtms - one per LTM, using non-floating self-ip
3) you can create gtm on one of the LTMs.
