Indexing Fundamentals


First, lets understand what the types of indexes are. If you take a reference book, you have the two types of indexes. The book itself is a clustered index, it contains all the data and has the key field of page number. The index in the back is your nonclustered index, it’s in order by the key field of subject, then has the implied second key field in there for page number, which nonclustered indexes in SQL Server have as well. On top of that, the table of contents at the beginning of the book is just like a b-tree built on every index that helps you get to the right page faster.

The key fields I mentioned are the fields everything is ordered by, just as you would expect a book to be ordered by page number. Nonclustered indexes also have implied key fields, which are all of the clustered index’s key fields that weren’t explicitly made key fields already, which is a great reason to keep your clustered indexes as narrow as possible in most cases, which means they have fewer columns.

If you have a phone book then you have a clustered index with the key fields last_name, first_name, address, phone_number. That means it’s ordered by each one, so any duplicate last names are still in order by first name, and so on. It’s wide for a clustered index, but doesn’t have any nonclustered indexes to worry about, so wide is fine here.

However, the important thing to notice is that if I asked you for people with the last name “Hood”, you’d find them rather quickly. If I asked you whose phone number is 412-555-6789, you’d cringe. The same thing happens in SQL Server where it will do the work you ask of it, but it won’t be cheap, quick and easy. This is why the internet has reverse phone number lookups where they have a nonclustered index on phone number which lets you find the name quickly by just knowing the number.

Indexes can also have included columns which are stored in the index, but no sorting is done on them. These fields don’t even exist in the b-tree for finding the information, only the page level where everything is stored. To simplify the process, this is where you put fields that you have in your SELECT clause but not your WHERE clause since don’t need it sorted to search quicker, you just need it to fulfill your query’s needs.

How are they used, and what do they cost?

A nonclustered index is used exactly the same way you read a reference book. You look up the key field in the back of the book and see what information is there. If you covered all the information that you needed just by looking at that index then you’re done. However, if you want to know more that wasn’t listed in the index then you take the key field of the clustered index (the page number) and do a key lookup by flipping to those pages and reading more about the subject at hand.

If you noticed, reading all that information in the back of the book seems pretty simple and efficient, especially when all of the information is there so you don’t have to touch the book itself. If you don’t have to touch the book then that’s called a covering index, which can be accomplished by adding the fields you want to retrieve to an index as either included or key columns. Depending on how many of these key lookups you have to do and how much extra information it has to get from the clustered index, it may make sense to make your index a covering index.

However, none of this is free. Every index you make has to be maintained. If you ignore filtered indexes, which I’ll get to soon, every time you insert or delete rows on a table then you have to write to each index. So if you have 6 nonclustered indexes, it doesn’t matter if you have a clustered index or a heap (merely a lack of a clustered index), you have to do one write to the table itself and one write to each nonclustered indexes, so 7 writes for that one row.

Add on top of that maintaining those indexes for fragmentation, extra storage needed on disks and backups, and more data that can possibly be held in the cache, and this stuff really starts to get expensive.

How many indexes should I have?

There’s no magic number of indexes you should have on any single table. Too few makes it hard to find the data, and too many makes a maintenance nightmare with data modifications taking longer. What indexes and how many will vary wildly, but there are methods to figure out what’s appropriate for each table. Chances are, many of your tables will benefit from both dropping indexes and creating them.

Consolidate what you have

First, look at the indexes themselves. If you have indexes that have the same first key field as the clustered index then you can consider removing those indexes with little to no detrimental effects. This is regardless as to if these indexes are being used or not.

Now, look at what indexes are never used. This can be done by looking at sys.dm_db_index_usage_stats to get you up to a month of data (assuming a monthly reboot for patching), or, my preferred method, by querying your tables you have set up that monitor this DMV. If an index is never being used it means it’s never helping you, but it is extra weight that’s adding up and slowing your database down. Just make sure you have enough data to say that the index really isn’t needed before relying on this alone.

Also, look for patterns in the indexes themselves by comparing the first couple key fields. If the couple key fields are the same, these are candidates for having the indexes combined into a single index. Depending on how quickly those key fields make the data semi-unique, you may be talking one to three key fields matching before I’m itching to reduce the number of indexes. You can do this by taking the columns that are in the second index that aren’t in the first and adding them to the first index as included columns, then drop the second index. Queries that used to hit either one of these indexes should run slower by an unnoticeable amount, yet you’re only storing data, doing updates, and maintaining a single index instead of two.

When you’re going through this process you just have to be careful that you understand the implementations of dropping a unique index. It is rarely a good idea to get rid of a unique index unless that uniqueness is enforced by other means as well.

Take this example.  I have a heap with a unique index on Col1 that gets scanned, but no seeks.  Also, there’s a NC index on Col2 that has a lot of seeks then is obviously doing a lot of lookups to get more information from the heap.  In this case I happen to know that the “unique” part of the index on Col1 is because that’s an identity column, so I’m not too concerned about maintaining the unique constraint here.  After seeing the usage stats over a month I’m able to restructure this table.  Both nonclustered indexes were dropped, a new unique clustered index was created on Col4 and Col1.  User Lookups were eliminated since it’s all just a clustered index now, we saved about 700 MB on disk, and the compressed equivalent on backups as well.

Heap_With_Unique_NC_Identity

Take another example where you can see that an index is duplicated.  No, none of these are 100% the same, but the second index is the unique (is_unique field), clustered (* for included columns, it’s just my way of doing it) has a single key column on Col1, which is the same as the first nonclustered index.  Seeing the usage like this, I’m going to drop the first index, saving 435 MB of space in my database.  This will have a negative impact in that it will move 626 index scans done throughout the month to an index that’s twice the size, and that’s a problem I’ll revisit later when I’m tuning queries that have high physical reads.

Notice you can also see two issues on this table that I’ll need to look into.  First, I have a nonclustered index that’s larger than my clustered index, something is wrong here.  Perhaps I should have put the fill factor on here to show that wasn’t the issue.  The real answer is that I’m working on a table that somehow got missed on the index maintenance, and I now know that this is possible and needs to be fixed.  Another issue is that there is a filtered index on here that has about 0.1% fewer rows than the unfiltered indexes, which means that you’re getting no benefits while alienating queries that may have used this otherwise.

Unused NC2

And one last example.  The 2nd and 3rd indexes both have a single key field on Col2 and just vary on the included columns.  This is a simple task of combining the two indexes by making one of them include Col1, Col2, Col6, Col7, Col8, and Col9, then dropping the other.  That saves us about 200 MB on disk, and avoids us having duplicated data in the cache.  Ask a SAN or backup admin and they’ll tell you the disk savings is great, but ask an unrealistically informed user and you’ll realize that cache savings is what makes a difference dropping your physical reads, upping your PLE, and making your app runs faster.

Then, there’s another simple task on this table of dropping the final index, which is nothing more than 112 MB of dead weight that was being updated an average of 30 times a day.  Not much to say here, they obviously didn’t use this as much as they thought they would.

Finally, you’re left with the info that User_Lookups is almost as high as the total number of seeks and scans on the nonclustered indexes.  This means that almost every query that hits the nonclustered indexes need to do a key lookup on the clustered index.  Knowing that, if I was going to get to that level of tuning, I would be looking into the queries that hit this table and determine if I want to add more included columns to actually cover the query or admit that I’ll never cover it efficiently and remove some included columns.  My answer here will vary partially depending on how many records are involved – key lookups are ok for 10 rows, but not so much for 1,000,000 rows.

Duplicated Index

None of these are hypothetical situations I forced AdventureWorks to do.  This is a real, production database, and I wrote two queries against it.  One to help me find these problems so I can put a change request in to fix them, and a second query that does the exact same thing while obfuscating the names to protect my employment.  So when you see the duplicated indexes or the sizes I’m working with, you can’t say that would never happen in production….it just did.

You reduced the load, now add to it

Now that you have fewer indexes in your database, it’s time to add to them. There are several ways to figure out what indexes you want to add, and I prefer to do everything manually. That means there’s no DTA (Database Tuning Advisor) here, it’s all scripts with proof that what I’m doing is making a difference.

Where are you finding the queries to tune

There are a couple approaches to finding the queries in most need of attention. Query Stats will give you the queries that take the most resources during your peak times, which will reduce the load on your server the quickest. Traces on duration will give you the queries that take the longest along with sample calls with all the parameters you need. Blocking monitors will give you the queries other queries are waiting on the most. Then, one of my favorites, job shadowing gives you what the pain points are for the users.

The first thing I look at is blocking, since that silently sits in the background making queries that could run extremely fast sit there and wait for a single slow query. If there is a lot of consistent blocking then take the head blockers and start tuning them. This won’t eliminate the fact that they’re grabbing locks, but it will cause those locks to be released quicker. Locks by themselves are not an issue and shouldn’t be eliminated, but the impact the locks have on other queries are an issue that is best to be handled this way. If you don’t have many locks that go too long, you can skip this and revisit it later. I say “too long” instead of giving you a number because it varies a lot; 5 seconds is too long for a web app while a minute isn’t much for a data warehouse.

Once that is out of the way, I’ll start diving into the most expensive queries on the server. They may not be the ones with the longest average run-time, but they add up to the most on the server. If you can cut the reads from 100,000 to 1,000 on a proc called every second then you’re doing more for the server than you would by tuning an hourly query that does 10,000,000 reads. Also, focusing on physical reads will help raise your PLE if that’s a concern on your server, which it usually is. Don’t stick on this for too long, you’re just trying to get the easy tuning out of the way and hopefully change the look of your Top-10 most expensive queries list a bit.

Now I move on to looking at the traces to see if any huge queries are running that are easy tuning opportunities. Again, skim through it, pick up the low hanging fruit and move on. Right now you’re still in the phase of reducing the load on the server, but you may not have hit what the users see yet.

Another place to look is in the cache to find out what indexes are taking up the most space there. If you know your data well enough then you should know what to expect from your cache. What you see there will probably surprise you, and you’ll probably see an Index_id of 0 or 1 (Heap or Clustered Index) sitting there taking 10 times what you would expect it to do. These are cases where there are clustered index scans taking place that you need to find. I was recently able to clear up 12 GB of cache on a server doing a full index scan on a 12 GB table where a 10 MB filtered index was able to do the job, which made PLE almost double.

So, now that you have the server running a little (or a lot) better in general, I get into the work that never ends. Get a hold of the people using the app, either by yourself or asking customer service to arrange it, and find out what they hate the most about the app’s performance. What you need here is for someone to write a list of when anything took a while, what screen they were on, and approximately how long it took. Don’t go requesting anyone to use stopwatches, you’re looking for “a couple seconds”, “about a minute”, or “I click OK then go get a cup of coffee.” Your goal here is to be able to correlate their pain points to something captured in your traces, giving you the procs they’re using along with real-life example calls to dive into.

Something you may find odd is that I’m diving into the part that affects the end users the most last, but I do it on purpose. The other pieces are things I skim through and do what I can in a week or so, but they’ll make the bigger impact on the server in general along with resolving some of the customer’s complaints in the process. In the process, you’re learning more about the server and how everything goes together so that when you start talking to your customers you can make a more immediate impact on their complaints.

After getting through list once I let my ADD take over, and man do I have ADD. The customers have an idea of what a DBA can do for them now to the point that they may even be contacting you directly when something annoys them, developers realize a SQL Server specialist can be a good person to lean on, and your server’s biggest pain point hurts a lot less. Once you get to this point, jump around on this list on your servers and take up a query or two when you feel like it.

Monitoring and Baselines Presentation


Here’s my full presentation for SQL Saturday #250 in Pittsburgh this past Saturday along with some notes on what I got out of it.

My Lessons

I learned a lot doing this, and I hope people learned a lot from it. For the people on the other side of the podium from me, the main lessons were in what I put in the abstract; they learned mostly about SQL Server. On my side of the podium though, the lessons were far from the same. Public speaking, creating presentation, prepping for unknown audiences, and seeing the gratitude of the SQL Server community in person. Anything that was on the abstract I learned slightly better, but that didn’t compare to what wasn’t on it.

Public speaking was always a fear of mine, and I tend to be someone who avoids my fears. However, throwing myself out there at the public as a whole was never my thing either, and I’ve been doing half decent, at least in my own mind. That being said, I decided to go all out and push myself a little further. After all, what are the chances of them picking me to present with all those professional teachers, consultants, and MVPs out there throwing up their abstracts. Best case I could say I tried, worst case I was going to throw myself out of my comfort zone and hope for the best.

They picked me, which I didn’t let myself expect. Everyone that knew I put in my abstract was also told by me that I only had a 50/50 chance at best, which was more optimistic than I really was, but I don’t like people knowing when I’m being dismissive of myself. It turns I was wrong a lot, and in every case I was wrong I was glad I was wrong.

Being Forced to Get Better

I’m not one to learn things for the sake of learning them, which is why I sucked in school. To really learn things I need a real-life use case, somewhere I’m going to apply it. I didn’t practice speaking ever before because I didn’t speak in front of large groups. I didn’t know how to put together a presentation properly because that’s not my thing…well, wasn’t is probably a better word now. Just like a couple months ago I wasn’t quite sure how to blog, but I’ve made it past that point.

Like everything in my life, blogging went from unknown to addiction quick. I’m not sure that I can say the same thing about speaking, but I can see the possibility. SQL Saturday only comes to Pittsburgh once a year and the local user group, which I’ll be joining soon, only meets once a month and has a single speaker. However, I can’t say that I didn’t look up when the SQL Saturdays in Cleveland and Washington DC were. Also, I took my own thoughts on my presentation and the feedback I received in and immediately starting thinking about how I could make that presentation better.

What’s below is the original, not touched up at all in the last 4 days. In part to show where I went wrong and how I’m going to fix it, and in part because Monday was my first day at a new job. By the way, starting a new job when between jobs is the best and worst thing you could ever do to yourself. You don’t have time to be too nervous about anything because you’re too overworked, but you’re also too overworked. Eh, you can’t win them all.

Well, on to the presentation. What you have here is my script that I talked through and taught myself before the presentation. However, if you were there, you’ll notice this isn’t exactly what I said. I didn’t read it, I presented it. There were no demos (the biggest complaint from the crowd) to avoid me from stumping myself, and the PowerPoint was just a whimsical picture for the start of each paragraph to keep me on track because I knew I’d be nervous and lose my way without a paragraph-by-paragraph guide helping me along.

I must have done at least descent because my reviews came back with two 3′s and the rest 4′s and 5′s out of 5. You’re just asking me to get off topic and start talking about why I hate ratings that are odd numbers, but no, I’m not getting off topic, I’m getting on to my presentation!!!

The Presentation

Download slide deck here.

I’ve been working on databases for over a decade, and most of that without having good standards or monitoring in place. My job was rough and I was much less effective. Now I know what I’m doing and I want to share that knowledge. In this presentation I am going to tell you what I watch and why I watch it. A presentation is a poor format to go through how to do this; having it in writing is much more beneficial to you. That is why I’m pretty much skipping that portion here and diverting you to my blog at SimpleSQLServer.com. Not so much to promote a blog that I lose money on, but instead to give you the resources in the best format.

As a DBA you will hit problems with the performance of your databases. It doesn’t matter if it’s one process or across the board, or if it’s just today or it has always been that way. No matter what it is, the more you know about your servers the easier it’s going to be to fix them.

It’s not easy, there’s no single spot to watch, no single solution to all of your problems, and there’s no “normal” values for these counters you can apply across every server you manage (Note: Thank you, Mike John, you stressed this point a lot). I watch several things on all the important servers, and most of these on every server. The important part is that you watch them continuously, even when you aren’t expecting to use the data. Some parts are cumulative and you can’t tell what was during an incident or what was from overnight maintenance. Other parts are snapshots and there’s no looking back.

(Note: Now I’m on to the stuff Brian Castle taught me. He’s the best you could hope for in a manager, and last I checked he was still hiring at EDMC in Robinson Township near Pittsburgh, PA)

To me, monitoring and baselining is the same thing. I know others will do a specific baselining process on a specific day and keep that. I feel you lose too much doing that and I watch enough to say that I have a continuous baseline going back for 13 months in most instances. I’m not all that worked up about lucky numbers or anything, it’s just nice to be able to say “that annual process we ran last week ran like this last year”. If you’re superstitious or have OCD, 400 days makes me just as happy.

Traces – Snapshot
Wait Stats – Cumulative – Resets on restart
Blocking – Snapshot
Query Stats – Cumulative – Resets on recompile
Index Stats – Cumulative – Resets on restart
OS Perf Counters – Varies
Database Sizes – Snapshot
Table Sizes – Snapshot

Traces – What long running processes have completed on the server?

You have to be careful, this one can kill your server. I have, on more than one occasion (sorry, Brian), caused a large server to reboot in the middle of the day because I filled the drives writing 40 GB of traces in 5 minutes. That being said, there is a safer way to trace, just don’t do an unfiltered trace catching everything.

I personally feel that absolutely every server should have a trace capturing RPC:Completed and SQL Batch:Completed that ran over X seconds duration. What is X? Well, that depends on the server. I’ve seen the best number for X be as low as 100 ms and as high as 10 seconds. Start high, and work your way down. You can add or do a separate trace for the Statement Completed for each of these if you want more detail.

You can get all of this except for SQL Batch:Completed with the text data from Extended Events, and that is a very good alternative. I haven’t made that jump for several reasons, although I would recommend it. First, this level of tracing hasn’t hurt me. Second, I still support several important servers that are still on SQL 2005 and I want to be as consistent as possible to provide as consistent as possible support.

What you’re trying to solve here is answering what ran long, have some hints as to why it ran long, and be able to say how often it has been running that way in recent history. A long duration doing very little work was waiting on something else, and other monitoring will help solve that problem. A lot of CPU, reads, or writes shows that you may need to look into tuning or statistics. Just keep in mind that reads can be reads of work tables, and writes can be writes to tempdb.

If you do it right you should have 4 or more days worth of history, and by doing it right I mean having 5 to 10 rollover files sized a reasonable size you can send off if needed (100 MB at most, they zip well) and capturing over the best duration threshold for your environment.

Wait Stats – What’s slowing you down?

The traces tell you what work was done, and the waist statistics tell you what went on when the query couldn’t actively do its work. This is cumulative over the server, so you can’t get details down to an individual process, however, you can see how much time is wasted and where it’s wasted at.

There are almost 500 distinct wait types in 2008 R2, but you don’t need to worry about that. The important thing is to know what are your biggest waits are, and you can look them up if you don’t recognize them. In the last slide I have links to a free eBook written by Jonathan Kehayias and Tom Kruger for the Accidental DBA that does an amazing job documenting what you can ignore and the meaning of the big ones you’re most likely to see.

If you’re looking for overall server health then you would want to look at waits over a long period of time, specifically the times of day you’d like to see better performance. For incidents, you want to look at what your wait stats are now and compare them to your running baseline. For me, the typical baseline I use is the same timeframe yesterday and 7 days ago. These are actual relevant days that have all but the changes you put in over the last couple of days, and it’s a true baseline for this specific server. If this server never waits on PageIOLatch, but it’s in your top 3 today then you instantly have a direction your heading in troubleshooting.

Blocking – Who’s in your way?

Locking occurs when one query is using data, and blocking occurs when another query needs to do something that is incompatible with that lock. If you don’t keep an eye on it then you’re looking back at a trace and seeing that a query took 1 second of CPU, did 1,000 reads, and no writes, yet it figured out a way to take 5 minutes in duration. The users complain the app is slow or unusable, and you’re giving them the answer that you’ll try to figure it out next time it happens and THEN try to find out the root cause.

That’s a bad idea. There are three types of users – those who don’t know you because things are running smooth, those that like you because you can say “I see exactly what happened and I can start working on avoiding reoccurrences”, and those users that you keep telling that you’re not sure what happened and you’ll try to get a better idea the next time it happens. I wish I could tell you that I could have more users not know your name, but watching blocking is a very easy way to get the users that like you to outnumber the users that hate you.

So, what can you do to tell them you know what just happened? My answer is to capture everything that’s blocking on the server every minute. Sure, there’s going to be a lot you capture that is actually only blocking for 10ms, and there are going to be things that were blocking for 59 seconds that were timed perfectly so you never saw it happen. The trick is to know that there is no perfect solution, and capture what you can. The DMVs are there to provide all the information, and being able to capture that on the fly along with in a proactive monitoring setup will make you look amazing to the users. If something blocked for 5 minutes, you have no excuse not to say “This query blocked this query, and this was the head blocker”. Sure you may have an excuse to say “It’s a vendor database and I passed it on to their support”, but at least you can tell the users something, which is always better than nothing.

Query Stats

The DMV dm_exec_query_stats is used by many DBAs, even if all we realized was that we were running one of the “Top 10 Most Expensive Query” scripts off the internet. The problem is that this is using it wrong. This DMV holds everything from when the query first went into cache and loses that information when the query goes out of the cache, even just for being recompiled. So the big process you ran overnight could still be in cache, leading you down a path that’s actually low priority for you. However, that relatively large process that runs every 5 minutes that just recompiled a couple minutes ago due to stats auto updating, it’s not even in there. This is too volatile to say that you’re getting good numbers out of it.

The answer isn’t to find somewhere else to get this information, it’s to capture this information more often and more intelligently. Capture it once and you have a snapshot of what it looked like, capture it again an hour later and you know what work has been done in that hour. Repeat the process over and over again, constantly adding another hour’s worth of data and saving off the latest snapshot in the process and you have some real information to go off of. This isn’t so much of what happened to be in cache at the time, it’s a pretty good summary of what ran and when it ran. Now you can query in detail, specifically saying “I want to know what queries are doing the most physical reads between 9:00 AM and 5:00 PM on weekdays so I can have the biggest impact when I tune this database.”

Index Stats

The DMV dm_exec_index_usage_stats isn’t too different from dm_exec_query_stats except that it’s cumulative since the time SQL services were started. Looking at a snapshot still leaves you half blind, unable to see what happened before the last time you rebooted the server and unable to tell when that index was used. Personally, seeing when it was used is less important to me than how much it was used long-term. Tracking this can do amazing things for you if you’re really fine-tuning a database.

This DMV is really under-used in my experience. As time goes on with a database the data changes and you add more indexes to make it run faster, but you don’t know if or when the index it used to use was being used by anything else so it stays there. Over time that means you have more and more unused indexes, and there’s no solid proof that they aren’t being used without you doing the work to collect that proof. This, in addition to duplicated indexes, adds to the workload for inserts and updates, adds to the workload of index maintenance, adds to the database size, and put additional strain on your cache and thus the PLE.

OS Performance Counters

This is one of the most misunderstood DMVs while also being one of the most useful. It’s giving you critical counters that give you an idea of how SQL Server is interacting with the hardware, but it’s doing it with several different types of counters that have to be measured differently. There are a couple great posts out on the internet to understanding how each one needs to be measured. As for right now, we’re focusing on why you want to watch it and what it will do for you. And wouldn’t you know that I saved the hardest one to describe for last, because each value is different and there will be controversy on which ones you should be watching and which ones are just a waste of your time monitoring.

PLE is one that this is no controversy about if you should watch it. This is the average age of the pages in cache, and a good measurement of when you’re doing too many physical reads. Peeking in on this from time to time is good, you can make sure it’s high enough. However, watching this will let you know when it’s dropping and help you dive into why it’s dropping. If you’re also running a trace you can see what did a lot of reads in that timeframe, and if you’re capturing Query Stats then you can find out which one of your large read queries is doing all the physical reads that destroy your PLE.

Server Target and Total memory are also great to watch even though they rarely change after the server is back up to speed following a restart of SQL services. The target memory is how much SQL Server would like to have, and Total memory is how much it actually does have. If target memory drops then there’s OS pressure you need to worry about. Also, as Total memory is increasing then it means the server is still filling the cache and PLE can pretty much be ignored. Of course it’s low, everything in there is new.

Page Lookups, Page Reads, Page Writes and Lazy Writes give you a better idea of how SQL Server is interacting with your cache and disks. If you want a real picture of what SQL Server is doing behind the scenes, this is much more useful than glancing at cache hit ratio.

Deadlocks and Memory Grants Pending are two things you’d like to always see at zero. That may be out of the question with deadlocks, but if you’re getting above zero on memory grants than you need to find out when that happened and everything that was running at that time.

SQL Compilations and Recompilations are hidden CPU hogs. They’re never going to be at zero because very few servers have stuff stay in cache forever, ad-hoc code runs, and several other reasons. However, when one of these counters jump up, you may have a hidden cause. For instance, if a piece of code that runs every minute or more was written in a way that it can’t be stored in cache, these numbers will be noticeably higher. When this happens, your CPU will be noticeably more stressed with no other indicators as to why, and this query won’t even show up in your Query Stats as it relies on showing the stats of everything that’s currently in cache.

Database Sizes

This seems simple, and it is. Most people know how to find the sizes of their databases, but do you track it? If not, you should. Some questions you should be able to answer readily are how fast your databases are growing, when will you run out of space, and is the steady decline in PLE justified by the growth of the database. Getting into more details, you can touch on when does the database grow such as an accounting app jumping in size every April, or answering if the data is growing steadily or exponentially. These are important things to know when it comes time to budget for new servers and disk space.

On the other side of things, you also have information to push back on application teams on how much space they’re using and if that’s really necessary. A lot of times you’ll find that they’re surprised by their growth to the point that you have to show them the numbers for them to believe it. It’s not uncommon for the response from this to be a cleanup project that helps keep the databases smaller and running faster, which is a big goal being accomplished.

Table Sizes

This is just building off of watching your database sizes in more detail. When a database starts filling up, and it’s filling up quick, it’s good to tell an application team what table is doing it. In my current environment it makes sense to watch every table that is over .5% of the database size AND over 100 MB. This means that some databases we don’t watch anything and other databases we watch about the 20 biggest tables. The biggest point is that we’re not trying to watch everything while watching anything big enough to make a difference to us.

If a database is filling a lot faster than normal then a lot of the time there’s a process that isn’t running like it should. The biggest table in the database may be a rather static value, but the third biggest table in the database wasn’t even on your radar two weeks ago. The app teams love it when you can tell them that not only is the database growing out of control, but we also see the growth in table X which has been growing at 200 MB per day starting on the first Saturday of last month. By the way, wasn’t last Saturday the day you changed a couple procs around or implemented an upgrade?

Now you just graduated from “Hey, there’s a problem” to also include “and here’s a huge lead to finding the root cause.” It’s rare to use this information, but it’s lightweight to capture for something that gives you a heavyweight appearance when you reference this knowledge to other teams.

My Critiques

I’d love to hear what you think about this presentation. Before we get to that, here are the complaints from the most harsh person in the room when I was presenting…

I didn’t do enough to say what each thing was. Although I marked the presentation as intermediate, you don’t know who’s going to be there. In this case, I knew a couple people in the crowd, and they ranged from data analyst to database manager. It should have been presented in more of “If you’re not ready for intermediate then you have everything you need, but it will take some effort to keep up”.

My slides were lacking. I stick with the idea that there should be few words on the screen, but I took it too far. I admitted above that the slides were there more to keep me on track then it was to help the audience, which I probably needed for my first public speaking venue. However, next time I’ll be less nervous and more prepared to make it geared for the audience better. Instead of random pictures, I should have more of graphs and data that I can talk through to give the audience visuals.

Nothing was given to the audience, and there should have been something. To be fair, none of the presentations that I know of handed anything to the audience. However, I set my own bar, and I feel that people would like a sheet of paper with the outline on it with links to online sources. My blog, of course, because it shows how to grab everything. Also, an outside link for each item. This gives a physical reminder to turn this knowledge into action.

There was no demo, and that was the audience’s biggest complaint. While I’m still not sure that I would do a live demo on my second go around, I’m going to hold myself on the remark above about my slides having graphs and data. That would also drag me away from the podium to talk through the slides, making it a more dynamic and interesting presentation; that would really help the audience take things in.

Your Turn

Now I need you to do two things. Put this information to work in your environment is what will help you the most. What will help me the most is you commenting on this post both before and after you implement any of this so you can help me become a better presenter.

Thank you for taking the time to read this!

Indexes – Unused and Duplicates


Indexes aren’t free, and many databases end up with unused indexes. Every time you make any update to a table you will be updating the clustered index (I assume no heaps, because I hate heaps), and every index that has uses one of the columns that were updated. Inserts and Deletes affect every column and will affect every index, with exceptions for non-typical indexes. These updates cost a lot, to the point that I have several processes that disable certain nonclustered indexes, do all of the work, then rebuild the indexes afterwards.

The problem is that you can’t tell very easily if an index is being used or not. You can look at sys.dm_db_index_usage_stats, but that information is very limited in that gets wiped out on index rebuilds, SQL service restarts (server reboot included), and anytime you ask it to. That’s not mentioning that it also ignores obscure usage of the index. However, the obscure usage is more of something to keep in mind than it is something to base all of your decisions on.

From my point of view, there are two answers to this problem.  There isn’t a right and wrong answer, it’s more of a step 1 and step 2.

The long and hard one is to capture the values in sys.dm_db_index_usage_stats on a normal basis (daily, before index maintenance) and give yourself long term statistics that you can determine if an index’s usage justifies its cost. Some indexes are updated 1,000,000 times for two uses. You obviously don’t need to keep this index around all the time, but you’re still paranoid about what those uses were. Were they ad-hoc queries run by a developer, or were they year-end reports run by the CEO? You can get some idea if you look at your history and check what day it was run. I would want to make a rollback script, drop the index, and make a searchable document of indexes I dropped in case anyone complained about year-end reports not working.

That was the long answer, which is right and is not replaced by the short answer. The script below is the short answer. If all of the key fields in an index match, or at least the first three in a larger index, then there’s a good chance that one of the indexes isn’t required or they could be consolidated.

For instance, if you have two indexes that have the same two key fields in the same order but the second index also includes two columns, then you’ll probably see usage on both indexes (see the long-hard solution above). If one index isn’t in use in this situation, your job is easy. If both are used, then look into dropping the index that doesn’t include any columns. Why? Because a slightly larger index rarely increases the overhead too much, and the queries that used the first index aren’t usually impacted too much by using a slightly larger index that has the same foundation.

This area gets shakey on how you should handle it, and is really one of the few areas in SQL Server I consider more of an art. For instance, the script below will show you indexes that have 5 key columns where the first three match. You look and find that both indexes are being used, but users are complaining that updates are a bit slow, too. I’d lean towards condensing by taking the the last two key columns of one index and including them on the other, but which index gets dropped and how do you test it?

Do my uses of “look into”, “rarely”, “usually”, and “too much” seem uncertain to you? Welcome to indexing.

Here are the scripts I use for both the short and long answers mentioned.  The first one is for the short answer, giving you indexes that are for the most part duplicated.

DECLARE @KeyFieldMatches INT

SET @KeyFieldMatches = 3 --Number of key fields to match in order

IF object_id('tempdb..#IndexList') IS NOT NULL BEGIN
    DROP TABLE #IndexList
END

IF object_id('tempdb..#IndexListShort') IS NOT NULL BEGIN
    DROP TABLE #IndexListShort
END

CREATE TABLE #IndexList (
      object_id    bigint not null
    , table_name varchar(150) not null
    , index_id int not null
    , index_name varchar(150) not null
    , index_column int not null
    , included bit not null
    , column_name varchar(150) not null
    , index_type int not null
)

CREATE CLUSTERED INDEX IndexList_Clu ON #IndexList (object_id, index_id, index_column, included)

CREATE TABLE #IndexListShort (
      object_id bigint not null
    , table_name varchar(150) not null
    , index_id int not null
    , index_name varchar(150) not null
    , column_names_first_n varchar(450) not null
    , column_names_key varchar(4000) not null
    , column_names_included varchar(4000) not null
)

DECLARE @object_id bigint
DECLARE @index_id int
DECLARE @List varchar(4000)
DECLARE @ListIncl varchar(4000)
DECLARE @ListShort varchar(450)

INSERT INTO #IndexList
SELECT o.object_id
    , table_name = o.name
    , i.index_id
    , index_name = i.name
    , index_column = ic.index_column_id
    , included = ic.is_included_column
    , column_name = c.name
    , index_type = i.type
FROM sys.objects o
    INNER JOIN sys.indexes i ON o.object_id = i.object_id
    INNER JOIN sys.index_columns ic ON i.object_id = ic.object_id AND i.index_id = ic.index_id
    INNER JOIN sys.columns c ON ic.object_id = c.object_id AND ic.column_id = c.column_id

SET @object_id = (SELECT TOP 1 object_id FROM #IndexList)
SET @index_id = (SELECT TOP 1 index_id FROM #IndexList WHERE object_id = @object_id)

WHILE @object_id IS NOT NULL BEGIN
    SET @List = ''
    SET @ListIncl = ''
    SET @ListShort = ''

    SELECT @List = @List + COALESCE(i.column_name + ', ', '') FROM #IndexList i WHERE i.object_id = @object_id and i.index_id = @index_id and i.included = 0 ORDER BY i.index_column
    SELECT @List = substring(@List, 0, Len(@List))

    SELECT @ListIncl = @ListIncl + COALESCE(i.column_name + ', ', '') FROM #IndexList i WHERE i.object_id = @object_id and i.index_id = @index_id and i.included = 1 ORDER BY i.index_column
    SELECT @ListIncl = substring(@ListIncl, 0, Len(@ListIncl))

    SELECT @ListShort = @ListShort + COALESCE(i.column_name + ', ', '') FROM #IndexList i WHERE i.object_id = @object_id and i.index_id = @index_id and i.included = 0 and index_column <= @KeyFieldMatches ORDER BY i.index_column
    SELECT @ListShort = substring(@ListShort, 0, Len(@ListShort))

    INSERT INTO #IndexListShort 
    SELECT TOP 1 i.object_id
        , i.table_name
        , i.index_id
        , i.index_name
        , @ListShort
        , @List
        , CASE i.index_type --Clustered indexes include everything
            WHEN 1 THEN '*'
            ELSE @ListIncl
            END
    FROM #IndexList i
    WHERE i.object_id = @object_id and i.index_id = @index_id

    DELETE #IndexList WHERE object_id = @object_id and index_id = @index_id

    SET @object_id = (SELECT TOP 1 object_id FROM #IndexList)
    SET @index_id = (SELECT TOP 1 index_id FROM #IndexList WHERE object_id = @object_id)
END

SELECT table_name, index_name, column_names_key, column_names_included
FROM #IndexListShort i
WHERE EXISTS (SELECT * FROM #IndexListShort i2 WHERE i.object_id = i2.object_id AND i.column_names_first_n = i2.column_names_first_n GROUP BY object_id, column_names_first_n HAVING Count(*) > 1)
ORDER BY table_name, column_names_key

DROP TABLE #IndexList
DROP TABLE #IndexListShort

The long answer takes a bit more work to figure out. You need to store the information long-term, keep a snapshot of what it looked like most recently, and do all the calculations to make sure you have the right numbers. It’s really no different than what I do for other DMVs, so you’ll see this process duplicated a lot if you follow my work. The tables for storing this information:

SET ANSI_NULLS ON
GO

SET QUOTED_IDENTIFIER ON
GO

CREATE TABLE [dbo].[DMVIndex_Usage_Stats](
    [StartTime] [smalldatetime] NOT NULL,
    [EndTime] [smalldatetime] NOT NULL,
    [database_id] [smallint] NOT NULL,
    [object_id] [int] NOT NULL,
    [index_id] [int] NOT NULL,
    [user_seeks] [bigint] NOT NULL,
    [user_scans] [bigint] NOT NULL,
    [user_lookups] [bigint] NOT NULL,
    [user_updates] [bigint] NOT NULL,
    [system_seeks] [bigint] NOT NULL,
    [system_scans] [bigint] NOT NULL,
    [system_lookups] [bigint] NOT NULL,
    [system_updates] [bigint] NOT NULL,
 CONSTRAINT [PK_DMVIndex_Usage_Stats] PRIMARY KEY CLUSTERED 
(
    [StartTime] ASC,
    [database_id] ASC,
    [object_id] ASC,
    [index_id] ASC
)WITH (PAD_INDEX  = OFF, STATISTICS_NORECOMPUTE  = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS  = ON, ALLOW_PAGE_LOCKS  = ON, FILLFACTOR = 90) ON [PRIMARY]
) ON [PRIMARY]
GO

CREATE TABLE [dbo].[DMVIndex_Usage_Stats_Temp](
    [TS] [smalldatetime] NOT NULL,
    [database_id] [smallint] NOT NULL,
    [object_id] [int] NOT NULL,
    [index_id] [int] NOT NULL,
    [user_seeks] [bigint] NOT NULL,
    [user_scans] [bigint] NOT NULL,
    [user_lookups] [bigint] NOT NULL,
    [user_updates] [bigint] NOT NULL,
    [system_seeks] [bigint] NOT NULL,
    [system_scans] [bigint] NOT NULL,
    [system_lookups] [bigint] NOT NULL,
    [system_updates] [bigint] NOT NULL,
 CONSTRAINT [PK_DMVIndex_Usage_Stats_Temp] PRIMARY KEY CLUSTERED 
(
    [TS] ASC,
    [database_id] ASC,
    [object_id] ASC,
    [index_id] ASC
)WITH (PAD_INDEX  = OFF, STATISTICS_NORECOMPUTE  = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS  = ON, ALLOW_PAGE_LOCKS  = ON, FILLFACTOR = 90) ON [PRIMARY]
) ON [PRIMARY]

GO

And now the script that captures the information and does all of your calculations. You may have noticed that the tables themselves do NOT store the index or table names which was intentional. Name changes happen, tables and indexes get dropped and created, all of that is extra work for little benefit. Is the benefit worth it? I thought it wasn’t. I’m not expecting everyone to agree with me here because there are arguments going the other way. My advice is STEAL MY WORK!!! Don’t copy it, you need to steal it. Make it your own, learn it well enough that you can improve it and build upon it. At that point you’ll truly understand every bit of it and hopefully start up your own blog. I hear EvenSimplerSQLServer.com is available, but that’s subject to change. If you don’t want to go that far, then write your ideas in the comment box below. I change these posts from time to time, and this script was actually added months after the original post was made. Just don’t ask me why it wasn’t here in the first place, my answer is just….DOH!!!! Even saying I was only 6 days into blogging at the time I wrote the original doesn’t make that acceptable.

DECLARE @TS as SmallDateTime
DECLARE @Old_TS as SmallDateTime
DECLARE @Reboot_TS as SmallDateTime

SET @TS = GetDate()
SET @Reboot_TS = (SELECT create_date FROM sys.databases WHERE [name] = 'tempdb')

--If update wasn't already done this minute
  -- Delete all but the newest two updates - Only the newest will be used, so one spare is kept
  -- Add the current stats
  -- Calculate the usage between the last two and save in the permanent table
    -- If any of the values go down then the DMV reset for that index
      --It's a horrible join clause, but it makes everything null and uses the isnull(x,0)
      --AND N.user_seeks >= O.user_seeks will be hit if database is refreshed, detached, etc.
IF 0 = (SELECT Count(*) FROM Perf..DMVIndex_Usage_Stats_Temp WHERE TS = @TS) BEGIN
    DELETE Perf..DMVIndex_Usage_Stats_Temp 
    WHERE TS NOT IN 
        (SELECT DISTINCT TOP 2 TS 
         FROM Perf..DMVIndex_Usage_Stats_Temp 
         ORDER BY TS DESC)

    SET @Old_TS = --Newest TS in the temp table
        (SELECT TOP 1 TS
        FROM Perf..DMVIndex_Usage_Stats_Temp
        ORDER BY TS DESC)
        
    INSERT INTO Perf.dbo.DMVIndex_Usage_Stats_Temp
    SELECT @TS
        , database_id
        , object_id
        , index_id
        , user_seeks
        , user_scans
        , user_lookups
        , user_updates
        , system_seeks
        , system_scans
        , system_lookups
        , system_updates
    FROM sys.dm_db_index_usage_stats
    
    INSERT INTO Perf..DMVIndex_Usage_Stats
    SELECT isnull(@Old_TS, @Reboot_TS)
        , @TS
        , N.database_id
        , N.object_id
        , N.index_id
        , N.user_seeks - isnull(O.user_seeks, 0)
        , N.user_scans - isnull(O.user_scans, 0)
        , N.user_lookups - isnull(O.user_lookups, 0)
        , N.user_updates - isnull(O.user_updates, 0)
        , N.system_seeks - isnull(O.system_seeks, 0)
        , N.system_scans - isnull(O.system_scans, 0)
        , N.system_lookups - isnull(O.system_lookups, 0)
        , N.system_updates - isnull(O.system_updates, 0)
    FROM Perf..DMVIndex_Usage_Stats_Temp N --N = New, O = Old
        LEFT JOIN Perf..DMVIndex_Usage_Stats_Temp O ON O.TS = @Old_TS
            AND O.database_id = N.database_id
            AND O.object_id = N.object_id
            AND O.index_id = N.index_id
            AND N.user_seeks >= O.user_seeks
            AND N.user_scans >= O.user_scans
            AND N.user_lookups >= O.user_lookups
            AND N.user_updates >= O.user_updates
            AND N.system_seeks >= O.system_seeks
            AND N.system_scans >= O.system_scans
            AND N.system_lookups >= O.system_lookups
            AND N.system_updates >= O.system_updates
            AND @Reboot_TS < O.TS --They're reset on reboot, so don't compare if it was rebooted since last reading
    WHERE N.TS = @TS
END

Index Columns and Names


Don’t trust the name of an index to tell you what it is. The name is wrong, or at least that’s what you have to assume.

I’m a huge fan of index naming conventions, with my favorite being starting out by saying it’s an index (IDX) followed by the table name, then the key columns, the letters INCL if there are included columns, each included column listed (if reasonable, just do all or nothing for each index), a U or N denoting if it’s unique or not, then a C or N denoting if it’s clustered or not. However, there are too many times I’ve seen an index naming convention get me in trouble where the index IDX_TableName_Key1_Key2_Key3_INCL_Incl1_Incl2_U_N actually not have the column Key1 in it, and it wasn’t unique either.

My answer for that is to query the system tables to tell me what each index is. This is exactly the information I would like to see in a naming convention and pulled from a source that’s never wrong.

SELECT TableName = t.name
    , IndexName = i.name
    , i.is_unique
    , IndexType = i.type_desc
    , FileGroupName = d.name
    , KeyColumnList = substring((SELECT (', ' + c.name)
                            FROM sys.index_columns ic
                                INNER JOIN sys.columns c ON ic.object_id = c.object_id AND ic.column_id = c.column_id
                            WHERE i.object_id = ic.object_id AND i.index_id = ic.index_id
                                AND ic.is_included_column = 0
                            ORDER BY ic.key_ordinal
                            FOR XML PATH ('')
                            ), 3, 2000)
    , IncludedColumnList = substring((SELECT (', ' + c.name)
                            FROM sys.index_columns ic
                                INNER JOIN sys.columns c ON ic.object_id = c.object_id AND ic.column_id = c.column_id
                            WHERE i.object_id = ic.object_id AND i.index_id = ic.index_id
                                AND ic.is_included_column = 1
                            ORDER BY ic.key_ordinal
                            FOR XML PATH ('')
                            ), 3, 2000)
FROM sys.tables t
    INNER JOIN sys.indexes i ON t.object_id = i.object_id
    INNER JOIN sys.data_spaces d ON i.data_space_id = d.data_space_id
WHERE t.name LIKE '%'
    --AND (i.is_unique = 1 OR i.type_desc = 'Clustered')
ORDER BY 1, 2, 4, 3

You may say that’s a temporary one-off inqury the doesn’t fix anything, and it is. However, the permanent fix is very invasive, will void your support contracts, can cause damage, would cause pieces of future upgrades to fail, may not work if it generates a name that’s too long, and other minor details. Assuming you have a home-grown database and absolutely no query hints specifying an index anywhere in your code, have a dev environment recently refreshed from prod, and have looked into every other issue that I never even considered, do I have some code for you!!!

SELECT Command = '--DON''T RUN THIS WITHOUT FIRST LOOKING INTO THE CONSEQUENCES AND UPDATING YOUR RESUME'

UNION 
SELECT Command = 'EXEC sp_rename ''' + TableName + '.' + IndexName + ''', ''' + 'IX_' + TableName + ColumnList + case when len(IncludeList) > 3 then '_INCL' else '' end + ISNULL(IncludeList, '') + '_' + case is_unique when 1 then 'U' else 'N' end + '_' + left(IndexType COLLATE SQL_Latin1_General_CP1_CS_AS, 1) + ''' , ''INDEX'''
FROM (
        SELECT TableName = t.name
            , IndexName = i.name
            , i.is_unique
            , IndexType = i.type_desc
            , FileGroupName = d.name
            , ColumnList = substring((SELECT ('_' + c.name)
                                    FROM sys.index_columns ic
                                        INNER JOIN sys.columns c ON ic.object_id = c.object_id AND ic.column_id = c.column_id
                                    WHERE i.object_id = ic.object_id AND i.index_id = ic.index_id
                                        AND ic.is_included_column = 0
                                    ORDER BY ic.key_ordinal
                                    FOR XML PATH ('')
                                    ), 1, 2000)
            , IncludeList = substring((SELECT ('_' + c.name)
                                    FROM sys.index_columns ic
                                        INNER JOIN sys.columns c ON ic.object_id = c.object_id AND ic.column_id = c.column_id
                                    WHERE i.object_id = ic.object_id AND i.index_id = ic.index_id
                                        AND ic.is_included_column = 1
                                    ORDER BY ic.key_ordinal
                                    FOR XML PATH ('')
                                    ), 1, 2000)
        FROM sys.tables t
            INNER JOIN sys.indexes i ON t.object_id = i.object_id
            INNER JOIN sys.data_spaces d ON i.data_space_id = d.data_space_id
        WHERE t.name IN (
                  'Table_1'
                , 'Table_2')
    )x
ORDER BY 1