?
Solved

Real-time failover server without clustering?

Posted on 2009-02-10
16
Medium Priority
?
692 Views
Last Modified: 2012-05-06
We want to set up a real-time failover server for a critical SQL application that runs on a dedicated Windows 2003 server.  Our initial thought was to simply use SQL replication to another server, but then we learned that there are directories of accompanying application files that must be replicated as well.  We are aware of clustering, and third-party applications such as DoubleTake, but we are hoping that there may be a less expensive alternative.  Placing the files on a network storage device would work, but that still leaves a single point of failure.

What do the experts say?
0
Comment
Question by:mainelytrusses
  • 5
  • 2
  • 2
  • +2
12 Comments
 
LVL 38

Accepted Solution

by:
Jim P. earned 1000 total points
ID: 23617436
Clustering is really your only  option.

Even replication needs human intervention to get it back up.

You can get away with using Windows Enterprise 2003 and SQL Server Standard edition for an active passive cluster.

You only need to buy the licenses for the Active node.

As far as network storage device -- in general, it needs to be a SAN not a NAS. You need the speed to be at least 1 GBPS and direct attached.
0
 
LVL 15

Assisted Solution

by:MohammedU
MohammedU earned 500 total points
ID: 23617437
There are some options but nothing is cheap other than database mirroring which is not similar to the failover clustering but it will work without any manuall intervention.
There are some hardware replication options but they are very expensive too...
http://technet.microsoft.com/en-us/library/cc917680.aspx

For other than sql serves files and folders you have to loop hw replication or some file replication.
0
 
LVL 85

Assisted Solution

by:oBdA
oBdA earned 500 total points
ID: 23617561
If real-time failover would be trivial to implement, Microsoft wouldn't limit it to the Enterprise Edition and the use of certified hardware. Server clustering, especially when databases like SQL are concerned, is highly complex and has its price. (The bright spot, maybe, is that since SQL 2005, a two-node cluster is supported; SQL 2000 had to be Enterprise Edition for a cluster)
The (more or less) trivial Network Load Balancing, on the other hand, has found its way even into the Web Edition of W2k3 (it was only in Windows 2000 Advanced Server before).
It's just a matter of what you need; if a failure of the server over several hours costs a lot of money, or if you need so much uptime uptime that you can't afford to take the machine offline for patching, then you need to invest into a reliable failover solution.
0
Concerto's Cloud Advisory Services

Want to avoid the missteps to gaining all the benefits of the cloud? Learn more about the different assessment options from our Cloud Advisory team.

 
LVL 51

Expert Comment

by:Ted Bouskill
ID: 23629687
The previous experts are 100% correct.  Clustering is very complex technology, therefore it is expensive and it is your ONLY option for real-time failover on any platform.
0
 

Author Comment

by:mainelytrusses
ID: 23632480
Thank you for the responses.  It seems that I wasn't clear with my question.  I should have emphasized that it is the data that is the issue - not the hardware.  I need to ensure that no data is lost in the event of a hardware failure.  I understand that there will be some down time during recovery, but I want the data to be replicated in real time so that when replacement hardware is put into service, the data is current.  Restoring from backup, or shadow copy, means losing changes made since the last backup.  SQL replication does what I need regarding the SQL data - but is there a better way, and what options are there for additional non-SQL data?
0
 
LVL 85

Expert Comment

by:oBdA
ID: 23632598
Since you're looking for real-time failover of the SQL services as well as file storage, then you have the option of implementing MSCS, and use a mirrored SAN as shared storage.
The MSCS would take care of a node failure, the SAN mirroring of a storage failure (the infrastructure components would obviously have to be redundant, too).
0
 
LVL 38

Expert Comment

by:Jim P.
ID: 23632679
You may want to look at Falconstor -- it can do offsite replication of drives and realtime mirroring of your SAN.

Costs money, but then your point of failure become your SAN units.
0
 
LVL 51

Expert Comment

by:Ted Bouskill
ID: 23636031
I understand your requirements, however, replication is actually a serious problem because corruption in one database can propagate to the other instantly!  Plus there is a big performance loss.  Plus there is a latency lag with synchronization that can be equivalent to transaction log backups.  Do you use transaction log backups?  That can minimize data loss.

First off, if you want to minimize loss of data due to hardware failure you should put the databases on a RAID 6 or higher array.  With top quality hardware throughput is excellent.
0
 
LVL 51

Expert Comment

by:Ted Bouskill
ID: 24117914
Sometimes 'No' is the correct answer.
0
 
LVL 51

Expert Comment

by:Ted Bouskill
ID: 24117929
Sometimes 'No' is the correct answer.
0
 
LVL 51

Expert Comment

by:Ted Bouskill
ID: 24123428
Based on the requirement of maintaining data and application files http:#23617436 is the correct answer.
0
 

Author Comment

by:mainelytrusses
ID: 24146119
I accept as proposed by WhackAMod.  Thank you to all for input and participation.
0

Featured Post

Nothing ever in the clear!

This technical paper will help you implement VMware’s VM encryption as well as implement Veeam encryption which together will achieve the nothing ever in the clear goal. If a bad guy steals VMs, backups or traffic they get nothing.

Question has a verified solution.

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

One of the most important things in an application is the query performance. This article intends to give you good tips to improve the performance of your queries.
Unable to change the program that handles the scan event from a network attached Canon/Brother printer/scanner. This means you'll always have to choose which program handles this action, e.g. ControlCenter4 (in the case of a Brother).
The viewer will learn how to successfully create a multiboot device using the SARDU utility on Windows 7. Start the SARDU utility: Change the image directory to wherever you store your ISOs, this will prevent you from having 2 copies of an ISO wit…
How to fix incompatible JVM issue while installing Eclipse While installing Eclipse in windows, got one error like above and unable to proceed with the installation. This video describes how to successfully install Eclipse. How to solve incompa…

809 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