Tag Archives: DBA

Uncategorized

Powershell 2.0 Get-Member Enhancements

2 Comments

The other day I blogged on a cool way to alter database file properties.  And I promised I would film the solution for you.  Well, I got home that night and started filming but when I pulled up sqlps I was stopped dead in my tracks by something I couldn’t explain.  The methods to get and set were missing.  I checked my other box and they were there, so what happened?  The only difference was that the box I wrote the blog from was XP and the box I was filming from was Win7.

So I pinged a couple really smart guys as MS (and I cc’d Buck).  So after a couple of days we have an answer.  There’s a new feature in powershell 2.0 that hides the getter and setter methods from you by default in get-member.  They’re officially calling it an enhancement.

There are 2 new parameters that get-member supports that allow you to control what you see.  You can now use –view and –force.

-view takes the parameters Extended, Adapted, Base, All

-force doesn’t take any additional parameters.

However if you’re looking specifically for getter and setter methods, you’ll have to use –force because –view doesn’t display them.  Here’s what that’ll look like:

dir | gm –force

Now you can see everything.  It’s interesting though is that by defalt, PS loads all the methods, it just doesn’t show them to you.

Here’s the MSDN blog that explains this and so much more.

And I’d like to give a special thanks to Robert Hutchinson at MS for chasing this down for me while Buck and I sat on the sidelines helpless.

Uncategorized

Powershell to Alter Database File Properties

2 Comments

Man, I still just get so jazzed when I work with powershell. Every time I need to do something long and hard(that was for you Buck), I look in powershell and it lets me just lay back and enjoy the long and hard.

So this time we’ve got like 85 DB files for our main DB. I know that’s a lot but it’s been spread across so many LUNs for so long and it’s just grown and grown as it’s filled up the LUNs. Now we finally got a couple really large LUNs and we’re trying to consolidate the files. And now that we’ve moved most of them to a single LUN, we want to stop the autogrowth on most of them so they don’t grow anymore before we empty them and then drop them. Because what’s the point really in putting them all on a single LUN if you’re not going to get rid of some of them. So we definitely don’t want them to grow anymore before we can empty them.

The task at hand was to set all of these files (most of them really) to not autogrow. The choices were do it in the GUI, or write a T-SQL cursor to go through all the files and create an alter database command in dynamic SQL for each one. Neither of those is very attractive considering that there are so many files. Of the 2 of them though, the cursor is the more attractive solution. Then I thought of powershell. I hadn’t specifically done anything like this in powershell before, but I figured I’d be able to get some love. I’m not going to teach you powershell here, or explain the script too much. I’m just gonna give you the code and if you have any questions you’re free to write, but it’s pretty straightforward. I’ll be making a video on this tonight for those of you who like to see things in action.

PS SQLSERVER:\SQL\MYSQLBox\DEFAULT\Databases\MyDB\filegroups\data\files> gci | where{$_.Name -ne “MyDB_Data”} | %{$_.set_GrowthType(“None”); Alter();}

OK, so a quick glance at the important parts.

1. Make sure you’re in the right node. Notice I’m in data\files.
2. Here I’m filtering out the one file I don’t want to be effected. So the main data file I want to still be able to grow. I’ll talk about that more in a min.
3. Don’t forget to use the Alter() at the end. Usually whenever an object has an alter method it wants you to use it. If you don’t it’ll change it in cache in your powershell session but you won’t see the change persisted to the DB. So you’re not really making the change to the object in SQL, just in your powershell pipeline. You can tell if something has an alter method by doing this at the cmdline ‘gci | gm’. If there’s an alter method there, then use it.

OK, you don’t have to filter. If you find it too much trouble or if your files are named in such a way that it would be too hard to snipe a single file for some reason, then you can change them all and then just go back to the GUI to change the last one back to what it was.

So there you go. We have a nice 1-line script to alter as many files as you need. And you can clearly tack as many file changes on there as you need. And the best part about it is that the code doesn’t really change if you need to do another file operation. All you do is change the method call. Ahhh, the power of powershell.

Uncategorized

Deleting Production Data

While talking to a group of devs late last week the topic of deleting prod data came up.  Since I’m the big MVP of the group they wanted to know what the official word from on-high is.  The question is simple, and the answer is even simpler and frankly it’s one of those that was solved so long ago it just doesn’t hold any interest for me anymore.  I’ll hash it out here again though because I know there are plenty of you out there who haven’t solved this problem yet. 

The question is, what’s the best way to delete gazillions of rows out of a prod table.  The issue is with both concurrency  and rollback.  The problem is that if you do a huge delete in a single transaction you don’t leave any room for anyone else to access the table during that time.  So your delete can bring a DB to a complete halt if you don’t do it right.  As well, you put your system at risk should something go wrong and the transaction has to rollback.  So why would the transaction have to rollback?  Well, there are a couple reasons really.  The first, and most likely, is that it could fill up the log and with no more log the transaction can’t continue.  The other possibility is that the server or the service gets restarted because someone didn’t know you were doing a big operation.  And while I’m at it, there is one more logical possibility.  Someone could get impatient with the operation and stop it without realizing that it has to rollback.  This is actually a scenario that I’ve seen played out many times and the result is always the same.  The system has to wait for at least as long as it was going.  And what’s even better is when the DB doesn’t come back online right away, that same user will bounce the service again and again. 

And of course this counts for updates as well… in SQL2K this was accomplished like this:

SET RowCount 1000

While (1=1)

Begin

Begin Transaction

–Delete stmt here–

if @@RowCount = 0

Begin

   Commit Transaction

   Break

End

    Commit Transaction

End

However, this doesn’t work in SQL2K5 and above where you have to do something more like this:

set nocount on

while 1=1

begin

delete top(1000)

from table1

where conditions…

if @@rowcount = 0

end

 

So ok, that’s how you batch delete/update in prod.  Only under very controlled and specific circumstances should you ever do these types of ops in a single batch.

Uncategorized

Experts are Sharp

6 Comments

You know I was thinking just this morning about the last round of interviewing I did to find a new DBA at work.  And that of course got me thinking about some of the interviews I’ve done in the past.  There are a few that really stick out.  The ones that are sticking out right away are the ones who didn’t know anything and claimed that they had so much experience and were so good that they didn’t have to be bothered with memorizing every little thing anymore.

This astounds me because all the experts I know are really sharp and on top of their game.  So what these guys are telling me is that they’re so good they don’t have to demonstrate even the most basic knowledge of SQL because they’ve transcended above that?  If that’s the case then my mother’s 100x the DBA any of us will ever be because she doesn’t know the first thing about it.

I remember this one guy especially.  He claimed both on his resume and in person to be an expert in query tuning.  He said, I’ve never found anyone who’s my equal at tuning queries.  So armed with that bit of knowledge I set about quizzing him with the basics.  I mean after all, you have to just get the basics out of the way, right?  I asked him if he had ever worked with exec plans.  He said of course, you don’t tune queries without them.  I said, that’s what I think, but i just wanted to make sure we were on the same page.  And I then asked him how expert his knowledge was of exec plans.  He said he was a very deep expert and didn’t know anyone with his knowledge.  Wow, now I’m getting a little nervous, right?

So I started with the basics.  What’s the difference between an index scan and an index seek?  Well, I’m not sure the exact difference, but I know you want to get rid of one of them.  OK, which one?  I can’t remember.  Um, ok.

So what’s a bookmark lookup (this was back when SQL2K was stull ubiquitous)?  I’ve seen it before, but I’m not sure what it does.

We went back and forth like that a couple more times and I finally broke down and told him that there was no evidence that he had ever tuned a query because he didn’t even have basic knowledge of exec plans.  I asked him what he was basing his claim of being an expert on.  That’s when he let me have it.  Look, I’m an enterprise DBA and I don’t have to know every piddling definition you dig up out of BOL.  Maybe someday when you’re at the level I am you’ll understand.

Um… ok, I’d say we’re done, huh? 

So like I said, I was thinking about that this morning and while I can’t keep up with everything, and nobody can, I like to think that I’ve got a lot of the basics covered.  And the real experts certainly know their stuff.  Go ahead and see how many people would follow her if you asked Kalen how big a SQL page is and she couldn’t answer.  And how many people do you think would follow Paul Tripp if he couldn’t tell you what DBCC CheckDB() was for? 

It just doesn’t hold water.  So for those of you out there thinking you’re all the Pooh, go test yourself and see how much knowledge you really have.  You may find out you’re not as hot as you thought.

Uncategorized

Valid Opinions

A friend of mine was in a meeting yesterday and he called me right after because he was upset. Apparently he and someone else had conflicting opinions about how to do something that neither one of them had ever done before… so they were both essentially guessing.

He was mostly upset because he said the other guy’s opinion was just as valid as his and so far it’s brought the group to a standstill while they figure this out.

I told him that there’s no way that both opinions are equally valid. In some instances that may be the case, but in this one, I happen to know who the other guy is and I know that when it comes to SQL brains he definitely got the short end of the stick. So I’m saying it out-right: not all opinions are equally valid. My mother-in-law likes to put her 2-cents in when we’re sitting around talking computers. She doesn’t know the 1st thing about it, but she likes to give her opinion anyway. So when I ask Jen’s brother something, his answer is worth more to me than her mother’s. The same holds true here right? Because someone is in a group and was invited to a meeting that doesn’t mean that they have equal say in things.

Here’s another really good example… let’s say that there’s a weird recovery scenario or even corruption scenario that you come across. And you come up to me and Paul Tripp at a conference and ask us what our opinions are. I’d say that Paul’s opinion in this case is worth far more than mine. Sure, I’m a qualified DBA and I’ve been around the block more than a few times, but Paul WROTE CheckDB so I think he carries a little more weight than I do. Even if it’s something neither of us has heard of before, I’d still take his guess over mine.

So no, I’m not one of those who believes that everyone’s say is as equally important as everyone else’s. Hell, I don’t even believe that everyone should even have a say. Some people are just not qualified to give opinions on a topic. In a restaurant you don’t see the dishwasher getting a say in what specials go on the board, and in a dojo you don’t see white belts chiming in on black belt promotions. So why is it that everyone in IT thinks they deserve an equal say just because they were invited to the meeting?

Uncategorized

The Best DR Plan in Existence

2 Comments

Someone just wrote to tell me that they just had a major outage on their main DW server due to one of the teams upgrading a driver.  There was nothing wrong with what they did, so there’s no reason to believe it would have brought the entire system down.

All I can say is that they had the best DR plan in existence and I had to share it with you so you can mimic it in your own shops.  All you have to do is not take any system-level backups and don’t save any of your configs or any of your files, or document any of your processes.  Then you can just have everyone in the dept wish as hard as they can that nothing bad will ever happen.

It must be an excellent DR plan because it’s the same one that so many companies use.  And it costs next to nothing to implement, which is probably why it’s so popular.  And the real joy is that you can put it into practice almost as quickly as you can put it on paper.  It takes next to no approval from the brass either.

I really don’t see the big deal with DR though.  There’s so much time online and in magazines/books dedicated to this topic and I just don’t get it.  If so many companies already have this practically foolproof DR plan then what’s left to discuss?  I’ve been thinking about this for a while now and I can’t come up with a single situation where this wouldn’t apply. 

Watch my free SQL Server Tutorials at:
http://MidnightDBA.ITBookworm.com

Read my book reviews at:
www.ITBookworm.com

Blog Author of:
Database Underground – http://www.infoworld.com/blogs/sean-mccown

Follow my Twitter:

http://twitter.com/MidnightDBA

Uncategorized

More Questionable Code

1 Comment

A couple pieces of code came across my desk yesterday and they’re definitely the kind of thing you read about… if nowhere else, then here for sure.  This was code that a dev sent me that is to be deployed next week.

The first one could actually cause a problem so I’ll talk about it first.

I’ll write the pseudo-code because I don’t have the actual code in front of me right now.  It goes like this:

1.  Check if the Whale schema exists.

2.  If it does exist then drop it.

3.  Create the Whale schema.

Ok, so that’s exactly what the script does. Here, I’ll go ahead and attempt the code.  I didn’t look it up so it may have a slight error in it, but you get the idea.

If exists (select name from sys.schemas where name = ‘Whale’)

BEGIN

drop schema Whale

END;

Create schema Whale with authorization = ‘dbo’;

So what’s wrong with this code you ask?  It’s simple, here’s the logic.

First, if it exists and you drop it then why just turn around and recreate it in the next step?  What’s the point of that?  So if your goal is to create the schema then everything above the ‘create’ line is useless.  And I know what you’re thinking… so what?  What’s the big deal if a tiny query gets run during a deployment?  It’s not like it’s going to drag the system down.  Well, you’re right about that, but it could kill the deployment.  If that schema did exist and it actually contained objects, then the drop statement would fail until you put those objects somewhere else.  So with there being no code to check that objects exist inside, you could be dooming the deployment to an unnecessary error.  You could also say that you know that the schema doesn’t exist so there’s nothing to worry about.  That’s fine, then take out the check and the drop.  If you know it’s not there, then take it out.  It’s called having concise code and it’s something that we lack in this industry these days.  Let me illustrate this with something completely ridiculous that also doesn’t really effect performance.

Create table #t1 (col1 int)

truncate table #t1

truncate table #t1

truncate table #t1

truncate table #t1

truncate table #t1

Insert #t1 Select 1

Ok, so look at that code above.  I created a temp table and then truncated it 5x.  That’s not going to effect performance at all because there’s no data in it since it was just created and there are no pages.  Then I go ahead and insert a row.  I can’t think of anyone who would let this kind of thing go on in an SP, or in deployment code, but we’re expected to let useless checks stay in our scripts. 

This code was clearly scripted in the wizard so it’s not like the dev went out of his way to do this by hand, but it’s the mark of a fairly inexperience coder to let the wizard create scripts and not inspect what it’s giving you.

The other piece of code doesn’t really do any damage as much as it’s just annoying.  In fact, it’s really 2 different scripts.  They’re create view scripts and the first one reads something like this:

create view MyView

as

Select MyFavoriteTable.col1,

MyFavoriteTable.col2,

MyFavoriteTable.col3,

MyFavoriteTable.col4,

MyFavoriteTable.col5,

MyFavoriteTable.col6,

MyFavoriteTable.col7,

MyFavoriteTable.col8,

MyFavoriteTable.col9,

MyFavoriteTable.col10

from MyFavoriteTable as MyFavoriteTable

Ok, so even typing this in the blog pisses me off.  And again, it’s just a style thing, but this just drives me crazy.  What drives me crazy the most is that this dev doesn’t understand what aliases are for.  To alias a table with the same name as the table itself defeats the purpose of having the alias in the first place.  Here a much better alias would have been mft or mt or m.  Hell, you could even go as far as MyFT or something like that.  Here, I’ll play one of those out for you and you tell me which one you’d rather read.

Select mft.col1,

mft.col2,

mft.col3,

mft.col4,

mft.col5,

mft.col6,

mft.col7,

mft.col8,

mft.col9,

mft.col10

from MyFavoriteTable as mft

Forget reading, which one of those would you rather type?  It’s just more concise and easier to read.  I happen to know the dev who wrote this and his opinion is that the full table name makes it easier to read when you’ve got larger joins.  Personally, I’ve never known anyone to complain about short aliases before, and again, I honestly think this boils down to inexperience.  When I first started coding well over a decade ago, I used to need things to be presented to me in very specific ways too.  It was the only way I could read the code.  That kind of thing is much less important to me now that I have been doing it for a long time.  Why?  Because I know how to code.

So the second thing about this script that bugs me is the fact that he saw the need to alias a query with a single table in it.  Now you’re just being obstinate for no reason.  I know the argument though.  He probably would say that he did it for consistency.  The only problem with that is the other create view script he submitted didn’t have the same stupid aliasing structure, so where’s the argument now?

Ok, so this code was clearly brought to us as part of a code review so the next logical question is why don’t we just reject the code if it bothers us so badly?  Well, the answer is simple.  Like so many places, our code reviews are merely perfunctory process placeholders.  Our lead dev has made it very clear that he’s going to do whatever he likes no matter what we say.  We’ve rejected code plenty of times for really valid process or performance reasons and he always thumbs his nose up at us and pushes it into prod anyway.  I really don’t understand how he’s gotten such a superior attitude towards the DBAs, but it’s completely unwarranted.  He acts like we’re in this just to piss on his parade and make him look bad, but we’re really only trying to help.  But we don’t even bother anymore.  What’s the point?  He knows more than the rest of us put together so we don’t even bring stuff up anymore.

So that’s my rant for tonight.  Remember, use aliases wisely and be as consistent as you can.  And for everyone’s sake listen to what your DBAs are telling you for a change.  They’re really only trying to help.

Uncategorized

1st Night at PASS

Ok we arrived in Seattle for PASS only I found out tonight that I’m not supposed to call it PASS.  Apparently the PC name for PASS is the PASS Summit.  Oh well…

So we started at the Tap House wtih Allen White and his wife and then went to the convention center (I can still call it that, right?) to register.  There we met up with tons of other MVPs and a few people that Jen follows on twitter.  Would that be her fellow twits?

I was talking to Allen about how things appear to be getting better because there are a lot more MVPs here than there have been the past couple yrs.  I think this is going to be a good week.

We also did our first DBAs@Midnight with Allen.  It was really fun.  We got to talk about Oracle and making mistakes, and powershell, etc.  It was a fun talk.  I’ll upload it soon so check back and I’ll let you know when it’s up.

Watch my free SQL Server Tutorials at:
http://MidnightDBA.ITBookworm.com

Read my book reviews at:
www.ITBookworm.com

Blog Author of:
Database Underground – http://www.infoworld.com/blogs/sean-mccown

Follow my Twitter:

http://twitter.com/MidnightDBA

Uncategorized

Firehat Coding

You know, sometimes you’ve just gotta do what you’ve just gotta do. And today is another one of those days where I end up multitasking at home because I’m watching my youngest boy (almost 2yrs).

So let no one say that you can’t code and play with your kids at the same time.

Hence, this is what I look like today.

Uncategorized

The Clean House

This post was particularly inspired while cleaning my house today.  It’s easy, if you think you’ve got a clean house then just stand up on a ladder some time and change your perspective.  I think you’ll find there are lots of things about your clean house you don’t know.  And while nobody’s officially complaining about the condition of the house (and in fact, everyone thinks it’s in great shape), it clearly is in disarray when you shift to that new bird’s eye view.

The same is true for your DB.  I honestly can’t count the number of times I’ve been told that there was no reason to look at DB performance because nobody was complaining about it.  Then when I dig into it and look at things from a DBA angle, the system is actually abysmal.  So if performance is so bad then why isn’t anyone complaining?  Well there can be a couple reasons for that.  First of all the DB could just not be very busy comparatively so the users won’t notice the issues.  And second, they may have just gotten used to things running slower so again they don’t notice it.  I’ve actually seen that several times.  Things get slower and slower, or better yet, they just start out slow and nobody ever knows the difference.  And what’s even WORSE is that more often than not there’s some cornhole in the background telling them that Oracle would solve all their problems.  Listen, if you can’t write a simple process against mssql and make it run right do you really think you’ll be able to hit Oracle ok?

So I’ve gotten off topic a little… back on track.  The point of all this is that just because nobody’s complaining about performance that doesn’t mean that your system is performing.  I quite often find that companies have no idea what their systems can scale to and they just assume that they’ll be able to handle whatever they throw at them.  And it typically takes more forethought than most companies have to look at a system objectively and proactively fix problems.

So the next time you’re considering your system performance, stand on a ladder or crawl down on the floor and see how things look from there.

Watch my free SQL Server Tutorials at:
http://MidnightDBA.ITBookworm.com

Read my book reviews at:
www.ITBookworm.com

Blog Author of:
Database Underground – http://www.infoworld.com/blogs/sean-mccown

Follow my Twitter:

http://twitter.com/MidnightDBA