|
Home > Archive > Windows 2000 track general > June 2003 > Redundant Site Links
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
Redundant Site Links
|
|
| Tech Ranger 2003-06-03, 12:46 am |
| I have tried to get an answer to this before based on having read about redundant site links in books. I had reached the point of accepting that this was an error by the authors, but it has been raised again. I was watching the video for 70-219 in the cbtnuggets series. Dan Charboneau, the instructor, clearly states and portrays on the screen the configuration of 2 site links between 2 sites. He says that this is for redundancy. In his example, one connection between the 2 sites is a T1 and the other is an ISDN line. He creates 2 site links. He names one ISDN. The other one is the DEFAULTIPSITELINK. He assigns costs accordingly. My question is how in the world will Windows 2000 know which site link represents the T1 line and which one represents the ISDN connection? | |
| mrfixit 2003-06-03, 7:39 am |
| You raise an interesting question. My understanding is that T-1 is a persistent connection, while ISDN is a short term connection, (Dial-up).
ISDN
Integrated Services Digital Network (ISDN) technology makes it possible to offer telephone customers digital data and voice services using a single wire by dividing the capacity of the wire into separate channels. A basic rate ISDN line can offer speeds of up to 128 kilobits per second (Kbps) using two 64 Kbps channels. An ISDN line must be installed by the phone company at both the server site and the remote site. In most instances, ISDN is used for intermittent, dial-up connectivity rather than for a persistent or permanent connection.
T-Carrier Line
The leased line has traditionally been a fast, permanent alternative to dial-up remote access. In most instances, this has been in the form of a T-Carrier line, such as a T1 or fractional T1 line that transmits digital data at a maximum of 1.544 Mbps by using the telephone-switching network.
Any clearer? | |
| Tech Ranger 2003-06-03, 7:44 am |
| So you have 2 sites connected physically by these 2 technologies. You want to use the T1 whenever possible for replication traffic, so you create 2 site links and assign costs. Just because you name one link ISDN doesn't mean Win2K will know that that site link is a logical representation of your physical ISDN connection between the sites. To the best of my knowledge, I don't see any feature of AD whereby the OS associates site links with physical connections. | |
| Shiryu 2003-06-03, 7:50 am |
| Tech Ranger
You assign cost to each line | |
| Tech Ranger 2003-06-03, 7:54 am |
| I wish you could see the video.
If you draw 2 sites, draw 2 lines between the 2 sites. write T1 under one line, ISDN under the other one. Assuming you have done the work in AD, you now have 2 sites and 2 site links. You assign a cost of 100 to ISDN and 50 to T1. Great, now the KCC will prefer to use T1. Question, how does the KCC know that the site link you have named T1 is a representation of your leased line???? The only association that I see is the fact that you named it as such, but that is for your benefit. | |
|
| A.D. always chooses the connection on a per-cost basis, so the cheapter connection will always be used if it is available. It's on you to make sure that you've got them setup correctly with the proper costs assigned.
In your case, I'm always going to use the link called T1 as long as it is available because you have assigned it a lower cost. The link called T1 may not be a T1 line at all, A.D. doesn't know what the physical circuit connection is, it just knows to use the link called T1 because it has the lower cost assigned.
Don't know if that helped you or not.  | |
| Tech Ranger 2003-06-03, 9:07 am |
| I have been trying humbly to ask a question, but I guess what I am really doing is asserting that the books that I have read and Mr. Charboneau's video are absolutely INCORRECT about creating redundant AD site links between 2 sites. You can hook up as many physical links between 2 sites as your heart desires. But, all that Ad needs is ONE site link created between any 2 sites you wish to have replicate with each other. Dan is directly stating that redundant site links are useful in case one line goes down. Also, costing them according to preference causes the preffered physical link to be used. I am stating that site link costs come into play when you have multiple sites and multiple replication paths. Costs between any 2 sites is used by the KCC to cost out the cheapest path. But cost cannot be used as a way to distinguish between 2 physical technologies used to connect 2 physical sites.
If I am wrong, I wish someone would show me where. I am very much interested in undrstanding this. | |
|
| Honestly, I guess I just don't understand what you are getting at. Sure, AD only needs one site link to start replicating, let's say it's a T1 connection. And when that T1 line gets nailed/severed via a backhoe incident, you loose replication conectivity. If I had a redundant site link established as a fail-over in case the T went down, then replication would continue via that link(assuming the links costs are assigned correctly). You are stating that you can not create two site links paths between the same two sites? (Just trying to understand what you are getting at Tech)
For example, if you have multiple connectivity options between DCs in a site, such as an Ethernet connection and a RAS connection, you can set up a redundant site link.
In the case of an RAS and an Ethernet connection, obviously the Ethernet connection would be much faster under almost any circumstances. You then assign a site link cost to each connection. Of the available site links, Windows 2000 will automatically use whichever link is cheapest.
I find it hard that multiple resources are all incorrect.
As for the inner working of the "magic" that make it happen, I do not know. | |
| Tech Ranger 2003-06-03, 12:15 pm |
| Maybe if I present it this way:
Physical infrastructure:
2 buildings each running a Win2K domain.
Building A and Building B are connected by a T1 line and by a 256K line.
I create 2 site links in AD.
I name them Charlie and Freddy.
I assign Charlie a cost of 100.
I assign Freddy a cost of 50.
Are you saying that Windows will associate Freddy with the T1? | |
|
| Ok, I understand what you are asking. Basically, I have two WAN links between sites A and B, a T1 and a 256K line. How do I make AD understand or how does AD know that they are different site links? (Associate the T1 with Freddy and the 256k with Charlie).
You know that's a really good question because as far as I know AD does not. AD doesn't route in an IP sense. I believe that it simply sees the aggregate links from point A to point B as one big link, and you'd use your routing infrastructure to manage the physical links.
So I really think AD sees the main T1 site link and 256K "redundant" link as one aggregate link instead of two seperate links.
Anyone else have a take on this? | |
| Tech Ranger 2003-06-03, 3:57 pm |
| I am glad that you understand my question. If you could see the video, you would see the instructor (Dan) say that he has the 2 sites connected by T1 and ISDN. He wants a backup link in case the T1 fails. But he doesn't want the ISDN to be used unless the T1 is unavailable. So he creates 2 site links as I have described before. I agree with what you have said in your last post. Win2K doesn't know about or care about bandwidth. That is why it requires YOU to give it info on costs. As far as I can see there is no way AD is going to identify your 2 connections and associate these 2 links with them as seperate entities. I may very well be wrong, but I wish someone who knows the answer would elucidate. |
|
|
|
|