Network Topology Help

Posted by alicecan

on 17 Nov 2006 8:41 AM

Posted by Rog

on 18 Nov 2006 11:34 AM

Posted by dbudny

on 20 Nov 2006 7:29 AM

Posted by alicecan

on 20 Nov 2006 10:02 AM

Posted by jci1975

on 21 Nov 2006 9:43 AM

Hi Alice,
Some of the verbiage gets a little nuts in terms of what is actually happening.

With your desired plan, what you are essentially doing is "load balancing" the network services spread among three servers. This keeps any one server from bearing the full brunt of serving up to the client computers, which can pull a network to a crawl.

Dbudny is referencing what "fault tolerance" or failover is whereby the remote server serves as a backup in case the primary facility becomes incapacitated for whatever reason - read: Disaster Recovery.

This is actually exactly where we both are, we are now exporing the possibility of installing a third server at our remote facility 35 mi away as a disaster recovery component.

Regarding the tape backup, consider the software. Backup Exec has limitations on one tape device backing up multiple standalone servers...we are exploring that option also.

A software package we are starting to look at is a product from EMC called Replistor for real time data copying between facilities. Looks very promising and not to stupid expensive.

You are correct, your SBS server is your 'primary' domain controller although Microsoft doesn't really use that nomenclature anymore. More appropriately, and as a carryover to Win2000, your SBS server, as the 'first' installed server in your domain, is the Domain Master, meaning it holds all the authentication records, and replicates those records to member domain controllers. In that way, you do have some local fault tolerance whereby any other domain controller can pick up the ball and carry authentication duties.

We are currently running 1 server with SBS 2003 as our "PDC" and a second Server 2003 Standard box that serves as a Terminal Server and a file server and a member domain controller. We have about 50 client computers accessing the system and our system is fairly resiliant (though aging). Your setup should be fine for several hundred users, although the network infrastructure should be looked at closely in terms of bandwidth capacity (10BastT vs 100BaseT and full duplex abilities.)

HTH

Cheers,

Ronnie B IT Manager Johnston County Industries http://www.jcindustries.com

Posted by alicecan

on 22 Nov 2006 9:16 AM

Posted by jci1975

on 27 Nov 2006 6:51 AM

Ronnie B IT Manager Johnston County Industries http://www.jcindustries.com

Posted by alicecan

on 28 Nov 2006 1:47 PM

Posted by jci1975

on 29 Nov 2006 6:15 AM

Ronnie B IT Manager Johnston County Industries http://www.jcindustries.com

Posted by shipley.c

on 30 Nov 2006 10:01 AM

I'd like to add my voice to the quality of Jeff Middleton's swing migration kits. I've used them twice in the past, and the tech support he is willing to give out is excellent. He's very knowledgeable and willing to assist.

You are not going to get the fault-tolerance you are looking for, however. At best, with this setup, if you want failover you're only going to be able to get login services. When Rog was talking about how SBS does not play nicely with other boxes, this is exactly the type of thing he is talking about. Let me explain further.

The products bundled with SBS are unique to the SBS installation. This means that you are not allowed to move the Exchange or SQL products to other servers. They remain on the SBS server and are required to do so. You cannot effectively set up high-availability, automatic service failover in a cluster environment with SBS anywhere in the mix. So increasing yourself to 3 servers total, but keeping SBS, won't give you any benefit. The SBS box must still run Exchange and SQL, as well as all the other roles. In short, you're going beyond what SBS was designed to do.

I've implemented a solution recently that utilizes SBS as the main login/exchange box and has a Terminal Server (server 2) and an application server (server 3)... but there is no failover on services. If one of the boxes goes down, then we have to find a way to bring the box back up.

One solution you may want to look at is Virtualization of your servers. Say for instance you're using VMWare to virtualize your SBS server. Let's say somehow the hardware fails (maybe it got melted by a lightning strike *shrug*). Well, you can bring that virtual server image (from your external backup) back online on a different box in a matter of minutes and start providing services again while you work on fixing or replacing the hardware.

I argue that you don't need failover as its too expensive to maintain for 28 users at this time. In the future you may need failover. Currently what you need to be able to do is have a reasonable time for complete disaster recovery. If you can hone that plan down to 2-4 hours time, then you have an acceptable level of service to provide your users.

Chris Shipley
Nutmeg Consulting

Posted by alicecan

on 1 Dec 2006 1:22 PM

Posted by alicecan

on 1 Dec 2006 9:07 PM

Posted by shipley.c

on 2 Dec 2006 5:48 AM

Chris Shipley
Nutmeg Consulting

Posted by alicecan

on 2 Dec 2006 10:23 AM

Posted by dbudny

on 4 Dec 2006 7:23 AM

New here?

If you'd like to participate, join us.
If you're already a member, log in.
Search for a specific post?
Subscribe to RSS

New Post

Details

Rate This

33 Replies
0 Subscribers
Postedover 12 years ago

Join TechSoup

Network Topology Help

Network Topology Help

New here?