Forum Discussion

Quintious_74258's avatar
Icon for Nimbostratus rankNimbostratus
Nov 02, 2011

Architecture Design question regarding GTM's

Alright, so here's the deal: Customer is going to VPN data to us, we have apps that are going to work some magic on the insidand then spit what the customer wants back out to them. We're going for georedundancy, so we have GTM's sitting in boxes at 2 datacenters. We're going to utilize LTM's on the inside for the apps, which have 2 components - the servers that run the app itself, and the servers that host the databases this app will be querying in order to work its magic. The GTM's are greek to me, the LTM's only slightly less so. Just need to know if I'm on the right path from a design perspective.



So, the design path I've got in my head here is Customer site (VPN initiation) > cloud > our edge router > our firewall > which then branches off to the GTM that syncs up with our other datacenter > & firewall further branches to our core switch > which contains our active/standby LTM's, which will be handling load balancing the incoming data to our app servers in one pool and also load balancing the queries from the app servers to our database servers in another pool, all of which are also physically connected to the core switch. And then of course, the same setup at our other datacenter, which will sync data so as to provide the necessary redundancy.



Do I have the right vision in my head, or am I misunderstanding where the placement of the GTM needs to be and whether or with regards to the usage of the LTM's? It just "feels" like I'm missing a component here.


2 Replies

  • Hi Quintious,



    The simplest way to think of the GTM is as a DNS Server (because in essence that is exactly what it is), but it has more intelligence.



    There are several different ways to set them up (Delegate an entire DNS Zone to them - or delegate a sub-domain - Either way, the request hits the GTM and it will then make an "intelligent" decision on what response it wants to give the client.



    You can use Geo-Location to determine which Data Center is closer and send the client there, or you can do the more traditional Active / Standby. Either way, the response to the client is the IP Address of the LTM Virtual Server that you want the client to go to (just like DNS).



    The GTM's are also not limited to just doing this for LTM's. They can do this for all types of devices.



    I would recommend reading the following:


    Overview of the Global Traffic Manager


    Overview of BIG-IP GTM Topology records



    Hope this helps.