We just purchased the STM (2000 series) and I've been configuring it and wanted to get some opinions on the way to do something. We're planning on pointing one of our Critical internal services over to the STM. (There are some pretty significant speed improvements event though we won't be doing any actual load balancing.) In the event of a Disaster all our clients will have their connections point toward the Corporate STM. We'll have an STM at our DR site as well and I'm trying to figure out the best way to have the clients connect to the DR STM in the event of a Corp failure. The service behind the STM will still have the same IP so today it isn't an issue.
From what I've seen, the only real way to do it would be to use GLB internally and point the clients to a DNS address. I think this will work but I'm not exactly sure how to configure it. I want all requests to go to Corp unless it isn't available. Our DNS infrastructure is MS based.
I'd love to hear your ideas!
Solved! Go to Solution.
Documentation for libDNS.rts (the TrafficScript DNS library) has now been published here: libDNS.rts: Interrogating and managing DNS traffic in Stingray.
Firstly, welcome to Splash!
There are (as always) several ways to "shave the yak" (so to speak...)
I have written a GLB Deployment guide here: Deployment Guide - Global Load Balancing with Parallel DNS that might be helpful if you go down the GLB route.
You could also investigate running your STM's as a multi site cluster. There is great info in the manual in chapter 28 on Multi Site Manager (MSM) that might also give you what you are looking for..
Remember that MSM and GLB can be used to great effect together also!
Let us know how you go...
Thanks for the response! I read your GLB document and it helped me understand the config from the user manual better. I don't think it'll accomplish what I was looking for however. All of our traffic will be internal traffic. We've got a very flat DNS structure right now. I think the only way to really do it would be to have all DNS requests use the load balancers (which I'm not opposed to.)
What I need at a basic level is www.example.com to always resolve to 10.10.10.1 until 10.10.10.1 is down then resolve to 10.10.20.1 until 10.10.10.1 is back. With 10.10.10.1 being a traffic IP for a pair of STM's and 10.10.20.1 being the DR STM.
We have our STM's as a multi site cluster already and that has made the configuration many times easier, I love that feature!
I am glad the doc was useful!
FWIW, in my experience, the changes required to set up a parallel GLB zone are usually much less user impacting than changing the DNS settings for everything...
If you are happy with all DNS going via the STM's, it sounds like all you need is straight up GLB then.
I was attempting to set up a parallel deployment but since all the traffic was internal, I was running into some issues. I created a new domain (ha) and told it to use the STM's at name servers but once I do that, I can't add any host records in because it delegated the subdomain. (I'm not an AD DNS expert by any means.) It's almost like I'd have to create an additional subdomain to get the extra layer. (I have a feeling I'm missing something glaringly obvious.)
Edit: I've done some playing around and don't think it work the way I was hoping. After revisiting the DNS changes, it would take quite a while for us to get those in place so we've decided against that. I think our best approach right is now to have Site Recovery Manager make a DNS change in the event of a failure. Thanks for your help!
Glad that you found a solution - let us know how you go!
Is it possible to get the STM to respond to a DNS request if we're not actually sending it to any back end servers?
IE: We have 2 fqdn's that we want it to respond to.
That's the only thing that will be going through GLB dns wise. I'm wondering if dev1 could point towards Traffic Group IP 1 and dev2 could point toward Traffic Group IP 2 and the STM could just respond itself. Then be able to make a decision about which addresses to serve. (TGIP1 at Corp or TGIP1 at DR based on corp being up or not.)
STM support libDNS now, so this can be achieved. I am looking to see where we have documentation on how to set this up. Standby...
Would it matter that we're using the Virtual Appliance?
Also, is it possible to tell the DR site STM to raise a TGIP of the Corp STM on a different NIC if the Corp STM's aren't available?
Also, thank you for the replies!!
Q: Would it matter that we're using the Virtual Appliance?
Q: Also, is it possible to tell the DR site STM to raise a TGIP of the Corp STM on a different NIC if the Corp STM's aren't available?
A: Simple answer: yes, longer answer: it depends on your topology and design. We would need more info on the deployment.