2005 High Availability Options

  • Comments posted to this topic are about the item 2005 High Availability Options

  • got this right as an educated guess because mirroring is the industry accepted 'easier' to set up, and as the new kid on the block is seen as better.

    However as GUIs are available to set up both mirroring and log shipping I'm not sure there is anything in it especially when you start considering the choices to be made in mirroring about high performance or high safety, using SNAC and setting up a witness server.

    Log shipping remains the easiest to understand and troubleshoot and that goes along way to making it easier to administer. Its down to which method suits your recovery requirements best.

    I am also surprised almost as many people have answered transactional replication as log shipping. I would never use replication for DR. Its for scale-out. Too restrictive in that the DB has to fulfill certain design requirements and you dont get all changes made to the database copied.


  • I agree, depending on the circumstances I think one could make an argument that Database Mirroring and Log Shipping both require little administrative effort once they are set up. Both have GUI setups, and can be easily scripted if desired.

    The Redneck DBA

  • another reason mirroring is seen as 'easier' is that microsoft included that claim in their marketing to encourage people to upgrade.......

    or am I being cynical 🙂


  • I disagree with the answer, but I guess it depends on your definition of "disaster". Having to restore a database because of hardware failure is extremely rare. Much more common is restoring because of a bug in an application or someone accidentally doing something to the database (using a SQL statement that had bad side effects, for example) and the mirror would be in the same condition.

  • I'll add my reasons in here why I approved this question. I think it's good, and I see the arguments why log shipping might be seen a easy.

    However, getting rights straightened, and ensuring that files actually copy across and get restored is, IMHO, more of an effort than mirroring. Mirroring works very similar, but by working with immediate restores.

    I agree that it might not fit and Log shipping may help you recover from user errors, but only if you have a decent delay between primary and secondary. If you restore every 15 minutes, which means you can come up pretty quickly if there's an issue, you also only have a limited time for someone to let you know a mistake was made.

    From an admin perspective, mirroring is easier, I think. Not that it's better or more suited to any particular environment, but that it's easier to get a copy of your production db.

  • I don't agree that the mirroring can recover an exact copy of the database when there is a disaster. Mirroring uses log shipping. Some logs may not apply to the mirror server when disaster occurs in production server. This is why you need to use command ALTER DATABASE SET PARTNER FORCE_SERVICE_ALLOW_DATA_LOSS in certain case. My problem with the question is the word EXACT. Am I wrong?

  • This seems highly subjective. In my experience, Log shipping is dead easy to setup and has virtually no maintenance overhead.

    [font="Times New Roman"]-- RBarryYoung[/font], [font="Times New Roman"] (302)375-0451[/font] blog: MovingSQL.com, Twitter: @RBarryYoung[font="Arial Black"]
    Proactive Performance Solutions, Inc.
    [font="Verdana"] "Performance is our middle name."[/font]

  • Following up a little more on my earlier comment...

    The hardware issue is also covered by disc mirroring on our production servers. The most common "disaster" I have had to deal with is logical corruption (the database is valid but has incorrect or missing data) and is generally resolved by going to a backup and reapplying recent changes with corrected code. DB mirroring makes sense but the scenarios in which it would be most useful rarely happen, IME.

  • I'd say it's a close tie between Mirroring and Log Shipping.

    If you wanted a fully HA Mirror solution then you would also need a witness server and an application that is modified to use the latest connectors allowing you to provide both the Primary and Failover servers. So that would mean maintaining to some extent 3 database servers, and assuming that you have applications using the latest and greatest db connection methods. But the failover would be automatic. So more work on the front end and more watching on daily maintenance for logs and 3 servers, but backend (ie. failover) should be transparent.

    Log Shipping has less work on the front end, little daily watching on maintenance, but a bit more work on the back end (ie. needing to use failover).

    So in the wash which one would take more Administrative effort?

  • I vote for Log shipping as well.

    I did selected "mirroring" because of all the hype around it but it is not IMO easier because of all the "new" concepts you need to handle and possible configuration options in terms of safety and failover. Log Shipping is very easy to understand and manage.

    * Noel

Viewing 11 posts - 1 through 11 (of 11 total)

You must be logged in to reply to this topic. Login to reply