| Author |
Topic |
|
Hannibal
Starting Member
20 Posts |
Posted - 2002-06-07 : 14:29:27
|
| Is it possible to cluster two SQL Server 2000 databases that are 1000+ miles apart? If so, how? If not, any alternative? |
|
|
setbasedisthetruepath
Used SQL Salesman
992 Posts |
Posted - 2002-06-07 : 15:01:52
|
quote: Is it possible to cluster two SQL Server 2000 databases that are 1000+ miles apart? If so, how? If not, any alternative?
No, you're 0/2. "Clustering" is attaching two or more servers physically to a shared storage device, along w/ MSCS to coordinate them. Microsoft does not recommend you run SQL Server with network attached storage, so the practical upshot is all servers involved need to be in the same rack.You might however consider log shipping, if your WAN is fast enough.setBasedIsTheTruepath<O> |
 |
|
|
Hannibal
Starting Member
20 Posts |
Posted - 2002-06-07 : 15:10:13
|
| Is it realistic to use log shipping every 0.01 seconds? :)The reason I asked about clustering from such a distance is because I came across an article about W2K Clustering stating it's possible (not based on distance, but instead latency).Thanks for the info, though. |
 |
|
|
bm1000
Starting Member
37 Posts |
Posted - 2002-06-07 : 18:52:17
|
| Log shipping refers to copying transaction log backups from one server to another and then restoring them. You would not do this 100 times a second. Once every 15 minutes is more like it. If you have a lot of money, EMC will sell you a mirrored SAN Solution. |
 |
|
|
Hannibal
Starting Member
20 Posts |
Posted - 2002-06-07 : 23:07:01
|
| bm1000, I realize that ... which is why I included the smiley after the comment. But thanks. ;)Does anyone have experience with this product? http://www.nsisoftware.com/main/pages/Products/GCspec.htmlIt says it can do it (clustering with servers in different locations), but can it actually do it? |
 |
|
|
robvolk
Most Valuable Yak
15732 Posts |
Posted - 2002-06-07 : 23:53:09
|
| How about a little more detail on your environment, from both the SQL Server end and the end-user side. How many people will be using the servers simultaneously? What kind of applications will be accessing the data? What performance requirements are there (meaning, how slow can it go and still be acceptable) Specifically, why do you need to cluster vs. having a standby server? Why won't replication work for you?Your situation is not one that is conducive to clustering, and I don't think you should continue in that direction unless you've totally eliminated every other possibility. Don't spend a lot of time or money on hardware and software, because you'll end up with a less than ideal setup. If the machines cannot be physically near each other, clustering will be more frustrating than helpful. Just because it's possible does not mean it's a good idea. |
 |
|
|
Hannibal
Starting Member
20 Posts |
Posted - 2002-06-08 : 10:40:24
|
| robvolk, basically I have a client who is asking my advice about things I'm not an expert in, but I have very little information to work with myself. I do know that they are trying to cluster their environment to fix a problem that isn't hardware or load balancing related (which is what I thought clustering was for). They have a primary server that will lose it's internet connection for various reasons (maybe a satellite goes out of range), and they want to switch all users over to another server that is not in the same location (otherwise that server wouldn't be able to access the internet either) - once the primary server has a connection again, they want to switch all users back. Originally they had asked for a standby/query server which is why most of my original questions centered around replication/log shipping. However, that was a miscommunication. To compound the problem, many tables in the database don't have primary keys (see my other question). In short, it's a mess. I don't have exact numbers, but the average day sees about 500,000 transactions.I do agree "just because it's possible, doesn't mean it's a good idea," but I don't know what else to suggest to them (other than get better connectivity). It's been a headache, since I'm the type of person that likes to solve problems at the root instead of implementing sloppy work arounds. Which is exactly what I feel clustering this environment would be.I appreciate the help everyone has given me on this forum, though. :) |
 |
|
|
robvolk
Most Valuable Yak
15732 Posts |
Posted - 2002-06-08 : 11:10:48
|
| The best suggestion to make, for now, is what NOT to do, and clustering in that scenario is something that they should NOT do.Since both machines have intermittent internet connections, NEITHER of them should be used for SQL Server at all. Hey, just for laughs, ask these morons what they're gonna do if BOTH machines lose their connection!I know they won't like hearing that, but you're absolutely right, they need to fix their connectivity issues before they do anything else. There is no way this current system will work. If they can't or won't set up a proper in-house server and connection, you should recommend getting a hosted SQL Server. Those are the only options, period. Hosts are not that expensive and the good ones (like Orcsweb) have superb support. |
 |
|
|
Hannibal
Starting Member
20 Posts |
Posted - 2002-06-08 : 11:38:07
|
| I can't recommend a hosted server, because of license issues. They are currently licensed to run their software in a specific jurisdiction and changing that would probably be costly and very time consuming. In short, they are on an island and any host on that island will face the same connection problems.Btw, the guys I'm dealing with aren't morons, they simply have the misfortune of running a system designed and developed by someone else. :)I guess what I have to recommend is to bring the database design up to par so they can actually create a standby & query server for now. As for clustering, I'll recommend against it even though I don't have an alternative solution.If anyone has any ideas about alternative solutions to the situation described above, please feel free to post them. Once again, thanks for the help! |
 |
|
|
|