Win2K TS Question

[Voyager1]

This is not a MetaFrame question, but it is a terminal services question - so I think this is the right forum.

DNS Round-robin - what Microsoft calls load balancing (but we all know is really load distribution): Kind of sketchy as to how to set it up. I have 4 Win2K Terminal Server boxes that users will access a single application from. The company has decided not to use MetaFrame, so they are coming in via the RDP client.

Is setting up DNS Round Robin as simple as establishing several DNS records with the same name, but different IP addresses? Does anything else need to be done at the servers/clients/DNS servers?

I've always done LB in MetaFrame, so I could use a little advice here.

TIA -

- Bill

"You can get anything you want out of life, if you'll just help enough other people get what they want" - Zig Ziglar

 

[Jehova]

I had a headache getting their W2k load balancing to work. It is pretty much round robin, so it's not nearly as useful as MF's load balancing.

I gave each of my 4 servers (TERM1-4) two IP addresses, for example 10.2.20.121-124 then gave them all 10.2.20.120 as an additional IP.

Then created a static entry for a box named TERM to IP 10.2.20.120.

After that, the load balancing worked fine for TS. Great if everyone logs in once in the morning and stays on all day, but gets pretty out of whack fast with people getting on and off throughout the day.

 

[CitrixEngineer]

I didn't have much luck with Microsoft LB when I tried to use it in a Metaframe 1.8 environment - it seemed to break the ICA browser. I therefore did not bother with it in any of the XP environments I worked in.

Multi-homing Metaframe servers tends to confuse both types of browser - ICA/IMA, which starts listening on the wrong subnet - often unexpectedly!

However, I thought that Microsoft's "load-balancing" was a way of giving multiple machines the same IP address, as opposed to multiple IPs to groups of machines?

Microsoft make little distinction between clustering and this definition of load balancing - in fact, it would appear to be more of a fail-over mechanism.

Everything is handled at the NDIS level; client requests are shunted off in turn to cluster members by the LB NDIS drivers. If a server stops responding, requests are not sent to it any more, but get redirected to other hosts.

Round-robin is enabled on the DNS server, in its property sheet, under Properties, Advanced.

Nothing needs to be done at the client end, since it's all handled by the LB service.

That's about the extent of my knowledge of M$ LB, I'm afraid.

I hope it helps

 

[Voyager1]

Thanks for the comments.... - Bill

"You can get anything you want out of life, if you'll just help enough other people get what they want" - Zig Ziglar