• When SQL Server runs a query in parallel, each stream will complete it’s set of work and return its result which could cause your clustered-index-ordered data to appear out of order.
• When another query is already scanning the data when your SELECT statement is run, the data will be read out of the order you’re expecting as your query will piggy-back on the other scan, then come back to the beginning to gather all the data requested.
• If statistics are out-of-date on your clustered index, or SQL Server otherwise chooses to use a non-clustered index to scan the data, the order of the clustered index will not be reflected in the result set.

But don’t take my word for it, check out what Conor Cunningham has to say on the topic, including some sample code to prove the point.

Congratulations, guys!

Readers - if you’re attending the PASS Summit in Seattle, stop by the voting kiosks to cast your vote for the Board of Directors. This crew of professionals is responsible for coordinating volunteer efforts, directing the community, and making a difference in how we do our jobs as SQL Server DBAs. It really does impact you.

If you’re not attending the summit, keep your eyes posted in your email inbox for a voting survey too.

You can start playing with the wiki now by clicking on SQL Server Wiki at the top of the page.

If you have an MP3 player or a portable video player and you want to download our videos automatically, you can subscribe to our podcast feeds here:

• MP4 (Apple) Video Feed
• WMV (Microsoft) Video Feed
• MP3 Audio-Only Feed
• Zune One-Click Subscribe for Video
• Zune One-Click Subscribe for MP3

Take measuring log space, for example.  If I wanted to find out how much log space has been utilized, I would dash off a DBCC SQLPERF(LOGSPACE) statement.  But SQLPERF(LOGSPACE) only shows used/free space and not much more.  I might use DBCC LOGINFO to see how the active log “moved across the ldf file”, then examine the values columns such as “active” for more understanding.  Or I could go even more old-school, and look at the PerfMon counters or perhaps at the PerfMon counters exposed through the system table sysperfinfo.

However, I want to thank Tibor Karaszi and Gert Drapers for this great tip.  Simply use this query:

SELECT SUM([Log Record Length])

FROM ::fn_dblog(null, null)

WHERE…

The ::fn_dblog pseudotable provides a wonderful amount of information about the transaction log. Read more about this undocumented function at http://www.novicksoftware.com/udfOfWeek/Vol1/T-SQL-UDF-Volume-1-Number-27-fn_dblog.htm, or google ‘::fn_dblog’ for lots of other sources of information about it.

Thanks,

-Kevin

If you’re a good worker, you’re going to get a reputation.  People are going to bring you work.  If you’re a nice person, you’re going to want to say yes.  It feels great to say yes, and it feels great to make people happy.  Problem is, the better you get at working (no matter what the job is), the more people are going to want to give you things to do.  It really does never end - and that’s great.  It’s job security.

Being a great employee means saying no.

A lot.

Great employees have to say no several times a day.  As I write this, it’s eight AM, and I’ve already said no twice.  It’s not that I like saying no - it sucks - but there’s a lot of work that needs to get done, and I can’t do all of it.

Saying no means that when something really cool comes along at a moment’s notice, you can say yes.  If you’ve booked yourself chock full of deadlines and you have to work 50 hours a week just to do your regular stuff, then you don’t get to work on the fun spontaneous stuff that comes up out of nowhere.

Here’s how to determine when to say yes and when to say no:

• Is it a project you’d be proud to put on your resume?
• Is it a project your manager is going to be proud of?
• If it doesn’t get done, is it going to reflect poorly on you - or on somebody else?  (Hint: the “stakeholder” is the person who stands to gain or lose by the task’s success.)
• Can it be handed off to someone more junior than you?
• Is there someone else who would kill to get their hands on this work?  (Hint: the person who WANTS to do the work is the person who will do the best job on it, sometimes even if they’re not the most qualified, because they’ll devote the most love and attention to it instead of racing to get it done.)
• Exactly what have you got on your plate right now?  (If you can’t immediately point to a list of all your outstanding tasks, you need to say no until you CAN point to that list.  I suggest RememberTheMilk.com.)
• Does it play into your GTD 50k foot view?  (Does it map to something in your career plan?  More on that later…)

I have two computers running sql server, sqlserver01 and sqlserver01. On sqlserver01 I backup the database locally to C:\SQLBackup\backupfile.bak (db size: 150GB) then I copy and paste the *.bak file to sqlserver02 via the network. It finishes copying but when I try to restore it on sqlserver02 it fails. I have used tools like Eseutil from exchange but no luck. It seems like the bak file corrupts on the transfer.

What is the best way to copy over a 150GB *.bak file to another server then restore the *.bak file in a timely manner?
-end of question-
OK, to answer this I’ll need to actually throw out some other questions about this scenario.

Since this is to be timely, I would argue that this situation would benefit greatly from a tool that gives offers compressed backups. Compression can reduce the 150GB backup file down to 15-30GB, greatly improving copy speed.

But that does not address the real issue here. In fact, there should be no problem copying the 150GB file across the network to another location. It may be time consuming copying the file, depending on network speed, but this should work fine.

One question I have is if this is occurring while performing a normal Windows COPY or are if something like FTP is being used. If FTP, make sure a binary transfer is being performed.

Specific error messages weren’t given, so I would wonder if any are received or if the two instances are running the same version of SQL Server. Verifying that you can restore the database to the same instance from where it was created would be good to find out.  You can restore it to a new database name to avoid overwriting your original.

The other option available, again depending on network speed, is to back up the database to the network rather than to the local drive. A compressed backup would be a lot faster here as well.

If you want to, please provide some additional details of the error and your environments - Operation System versions and SQL Server versions.

Thanks.

–
David

Look around your office and find somebody who would love to have your job. If you’re a DBA, maybe it’s a developer who’s tired of learning new languages every 3-4 years and would like to stick with one (T-SQL) for a while.  Maybe it’s a network admin who’s tired of fixing desktops and servers and wants to make more money.  But either way, make sure it’s someone that you enjoy working with, because you’ll be training them for a while.

Approach them and start by saying, “Have you thought about becoming a DBA?  I’ve seen you in action, and I think you’d make a good DBA.  One of the toughest things about becoming a database administrator is getting your first job as a DBA, though, and I’d like to help you by getting you started with training.  One or two days a week, I’ll show you what I’m doing, and you can take over some of the basic tasks over time if you want.”

If they clap their hands and offer to wash your car, they’re your new junior DBA.  Even if YOU are a junior DBA, there’s somebody who wants this training, and they’ll be thankful for the opportunity.

If you’re paranoid and you’re worried that they’re going to steal your job, relax: we’re not going to give them the fun parts.  Give them the basic, day-to-day keep-the-lights-on tasks that you don’t particularly enjoy - things that won’t get you promoted.  Knowing how to check all of the backups isn’t going to make you famous, and every hour that you free up out of your schedule means an hour you can spend learning the latest & greatest tools and tricks.

Training this person will mean extra time initially on your part, so focus on training them on things that will pay off for you.  You don’t want to train them on something that you only do once a month, because it won’t free up your schedule and it won’t be useful to them either.  Train them on things you do all the time, like:

• Interpreting sp_who and sp_who2 to find out who’s blocking queries
• Reading query plans to improve a query
• Indexing tables or removing unnecessary indexes
• Backing up and restoring databases
• Installing the client tools

Or whatever else you find yourself doing more than once a week.

If this concept sounds crazy to you, ask yourself: how many times has your boss come up to you lately and said, “You’re doing a great job.  How would you like to take on this cool new tool we got in?”  Or is he handing you the crappy stuff that nobody wants?  If you’re getting the short end of the stick, or if you’re never getting to play with cool toys, it’s probably because your manager sees you as too overwhelmed with your current responsibilities - and only you can fix that.

“First and foremost, 32-bit is done. History. Archives. Windows Server 2008 R2 is the first Windows OS platform to go 64-bit only, and frankly it was high time. Customers have been unable to purchase a 32-bit server CPU for over two years now, and the advancements in CPU architectures really dictated that we squeeze as much performance out of customers’ hardware purchases as possible. The move to 64-bit is a first step.”

SQL Server database administrators everywhere are probably clapping with delight, but that clapping may slow down a little if you use virtual machines on your desktop computer for testing or development.  If you’d like to run Windows 2008 R2 as a virtual server guest, that means you need to be running a 64-bit host - which means your laptop or desktop will need to be running a 64-bit version of Windows.

Update: fellow Questie David Gugick points out that you can still run 64-bit guests on 32-bit host OS’s as long as your underlying hardware is 64-bit compliant.

I haven’t seen a good howto article that helps with SQL Server monitoring with MOM - until now. Tom LaRock, who moonlights as SQLServerPedia’s Monitoring Editor when he’s not fighting crime, wrote an in-depth article for Simple Talk:

Operations Manager: A Big Tinker Set