Solved

Windows 2000 Server and Server to Server Replication

Posted on 2004-09-15
10
461 Views
Last Modified: 2013-11-15
Hi there,
  I am looking for some suggestions on software that I could use in a corporate environment to have 2 Windows 2000 server's duplicate themselves completely!  What the corporation wants to experience is the ability to have complete fail-over on the SERVER.  To the point of, one SERVER dies, and in a matter of seconds the 2nd one takes over and from the user's point of view there was no change and they are still up and running while the 1st SERVER is corrected.  I would greately apprecate any experiences or maybe some links to software that can provide this solution.  

Thank you in advance!
0
Comment
Question by:Krocodile
  • 5
  • 3
  • 2
10 Comments
 
LVL 84

Expert Comment

by:oBdA
ID: 12070237
Depending on which services the "duplicate servers" should offer, Microsoft Clustering might be what you're looking for. Prepare to spend quite a bit of money for that, though.
Clustering Services
http://www.microsoft.com/windowsserver2003/technologies/clustering/default.mspx
0
 

Author Comment

by:Krocodile
ID: 12070378
I have looked at clustering before in the past, however, from what I have gathered clustering ususally requires an additional storage device that is somehow connected into the clustering environment in order for it to work correctly (i.e. like a NAS).  To be more specific about software I'm interested in.  I would like for the software to work directly on each SERVER, and not require any soft of additional hardware in between.  oBdA, maybe you could correct me if I'm mistaken about my understanding of clustering?

Thank you for your comment!
0
 
LVL 18

Expert Comment

by:exx1976
ID: 12075954
I have three clusters, and indeed, there is a need for shared storage between them.  We don't have a SAN (I don't think NAS would work), so we used direct-attached SCSI on a shared bus.  IBM makes some great stuff, and a file server with 1.2TB usable formatted space only cost me ~$25k...

IBM x345 servers
IBM 6M ServeRAID cards
IBM EXP400 external storage cabinet
throw in your drives..

For this, you get Ultra320 SCSI shared storage, and the servers failover immediately..  A complete failover/service start takes appriximately 90 seconds.

You may also want to do some homework on Windows 2003 clustering, since it allows the user of a local quorum drive (you must have at LEAST a quroum drive so that when one server takes over for the other it knows what's going on)..  Not sure, never tried it, but it might allow you to build a cluster without shared storage..

HTH,
exx
0
Best Practices: Disaster Recovery Testing

Besides backup, any IT division should have a disaster recovery plan. You will find a few tips below relating to the development of such a plan and to what issues one should pay special attention in the course of backup planning.

 
LVL 84

Accepted Solution

by:
oBdA earned 500 total points
ID: 12078251
Roughly spoken, a two node cluster as example: In a cluster, you can create (virtual) resources (IP addresses, names, file shares, services ...) that can be accessed by the clients. Only one of the machines actually owns the resource and offers it to the network, while the other monitors the first server (the situation of an active/passive cluster). The other nodes might as well offer other services, and be in turn monitored by the first (an active/active cluster). In case one node goes down, the resources are switched over to the other node, and the rest of the network can (mostly) move on and won't even notice it.
Whether you need external storage depends on your needs. In W2k3, you can theoretically do without (using the Majority Node Set feature for the quorum); but as soon as you want to have a clustered file server, there's no way around it. Your corporation won't get a working and reliable failover solution for the price of two standard servers.
The cheapest solution to test if this is what you're actually looking for is to download the evaluation copy of Virtual Server 2005; with this, you can setup two virtual machines as cluster nodes, without having to invest in hardware just for testing (assuming you have a machine with enough memory for Virtual Server).

Microsoft Virtual Server 2005
http://www.microsoft.com/windowsserversystem/virtualserver/default.mspx

Available Features in Windows Server 2003 Clusters
http://support.microsoft.com/?kbid=278007

Quorum Drive Configuration Information
http://support.microsoft.com/?kbid=280345

Network-Attached Storage and Server Cluster Support
http://support.microsoft.com/?kbid=812504

TechNet Support WebCast: Majority Node Set support in Microsoft Windows Server 2003 clusters
http://support.microsoft.com/?kbid=838612

Cluster Service Startup Options
http://support.microsoft.com/?kbid=258078
0
 

Author Comment

by:Krocodile
ID: 12086409
Thanks everybody for the wonderful suggestions and information!  I am going to do some reading from the provided links and definitely consider Clustering as being an option.  
0
 
LVL 84

Expert Comment

by:oBdA
ID: 12086551
One additional hint: The shared storage in the Virtual Server is based on a shared SCSI bus. You can of course do that in "real life" as well, but depending on your failsafe needs, this might not be enough. Shared SCSI has the big disadvantage (compared to a proper fibrechannel solution) that you won't be able to completely (physically) shut down one of the nodes while the other keeps up his business, as the SCSI bus won't be terminated anymore.
0
 
LVL 18

Expert Comment

by:exx1976
ID: 12086629
obda - I don't know where you got this idea from, but you most certainly CAN COMPLETELY shut down one of the nodes..  I have posted the hardware that I use above, and I have had to completely down a node on more than one occassion..


-exx
0
 
LVL 84

Expert Comment

by:oBdA
ID: 12087744
exx1976,
I've read that in some Microsoft document some time ago (that I can't find at the moment), where the setup of a shared SCSI bus for a cluster was described. They explained in detail that you should boot the other node and enter the BIOS or let it wait in the boot menu (especially during setup, when only one of them is running), instead of just turning it off. So unless the hardware tech who installs the stuff gives you a written statement that the SCSI bus is correctly terminated even if one node is completely taken off the power, I wouldn't trust it.
I don't doubt that it worked for you, but can you really guarantee, with a clear conscience, that for any given hardware setup, you "most certainly CAN COMPLETELY shutdown one of the nodes"?
0
 
LVL 18

Expert Comment

by:exx1976
ID: 12087845
I don't know about "any given hardware setup", but I know for 100% certain that I can guarantee it with a clear conscience for the hardware that I use.

And, the problem you describe, and the resolution (booting to BIOS), AFAIK, is not to remedy the SCSI termination issue (since SCSI busses have been self-terminating for quite some time now), it is to remedy the situation caused when the heartbeat cable thinks it's been unplugged.  You don't want this to happen for a prolonged period of time.  Not only that, but during the cluster build process, you can't have both machines connected to the SCSI device at the same time without corrupting the disk (before the cluster is initialized), but you need to be able to communicate with the other node during the setup..  The solution?  You guessed it, boot to the BIOS so the other network card shows as alive..

However, with the IBM servers I have (and I'm sure a lot of other models as well), the server doesn't even have to be on, it just has to have power, and the NIC will be active...

0
 
LVL 84

Expert Comment

by:oBdA
ID: 12089103
The article referenced specifically the SCSI termination; I'm not making this up.
The unloading of the TCP/IP stack when the network cable is unplugged can be prevented in W2k, btw.:
How to disable Media Sense for TCP/IP in Windows
http://support.microsoft.com/?kbid=239924
This is not necessary anymore in a W2k3 cluster; check the 278007 KB article from the links above.
Not to mention that crossover cables usually work, but the crossover function is no part of 802.3.
0

Featured Post

NFR key for Veeam Backup for Microsoft Office 365

Veeam is happy to provide a free NFR license (for 1 year, up to 10 users). This license allows for the non‑production use of Veeam Backup for Microsoft Office 365 in your home lab without any feature limitations.

Question has a verified solution.

If you are experiencing a similar issue, please ask a related question

Suggested Solutions

Storage devices are generally used to save the data or sometime transfer the data from one computer system to another system. However, sometimes user accidentally erased their important data from the Storage devices. Users have to know how data reco…
By default, Carbonite Server Backup manages your encryption key for you using Advanced Encryption Standard (AES) 128-bit encryption. If you choose to manage your private encryption key, your backups will be encrypted using AES 256-bit encryption.
This tutorial will walk an individual through the process of configuring basic necessities in order to use the 2010 version of Data Protection Manager. These include storage, agents, and protection jobs. Launch Data Protection Manager from the deskt…
This tutorial will show how to configure a new Backup Exec 2012 server and move an existing database to that server with the use of the BEUtility. Install Backup Exec 2012 on the new server and apply all of the latest hotfixes and service packs. The…

839 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question