Friday, January 18. 2008
What is PgAgent?
PgAgent is a basic scheduling agent that comes packaged with PgAdmin III (since pre-8.0 or so) and that can be managed by PgAdmin III. PgAdmin III is the database administration tool that comes packaged with PostgreSQL. For those familiar with unix/linux cronjobs and crontab structure, PgAgent's scheduling structure should look very familiar. For those familiar with using Microsoft SQL Server Scheduling Agent or Windows Scheduling Tasks, but not used to crontab structure, the PgAdmin III Job Agent interface to PgAgent should look very welcoming, but the schedule tab may look a little unfamiliar.
PgAgent can run both PostgreSQL stored functions and sql statements as well as OS shell commands and batch tasks.
Continue reading "Setting up PgAgent and Doing Scheduled Backups"
Wednesday, January 16. 2008
Recommended Books: MySQL High Performance Optimization, Backups, Replication
MySQL and Sun?
We just read that Sun is purchasing MySQL for a little under a billion. We are a little shocked and not quite sure what to make of it or how this affects Sun's investment in PostgreSQL. Further comments on the deal on Jignesh Shah's blog and Josh Berkus blogs. Jignesh and Josh both work at Sun and do PostgreSQL work as well.
Couple of random thoughts
Overall we think the move should prove positive for both camps.
PostgreSQL 8.3 really around the corner
8.3 is now on release candidate 8.3 RC1 and as Bruce Momjian noted, it looks like there might be an RC2.
We've been playing around with the 8.3 betas and RCs and really like the integrated Full Text Indexing and XML features. The new features make it possible to do a quickie REST service-based application. In the next issue of this journal, we hope to demonstrate creating REST services using 8.3 with server side - (PHP and/or ASP.NET) and front-end Adobe FLEX. We would have liked to demonstrate SilverLight/MoonLight as well, but we want to wait till Silverlight 2.0 hits release. We'll try to use the Pagila demo database for the upcoming demo app as Robert Treat has suggested.
Saturday, January 12. 2008
How do you delete duplicate rows in a table and still maintain one copy of the duplicate?Answer:
There are a couple of ways of doing this and approaches vary based on how big your table is, whether you have constraints in place, how programming intensive you want to go, whether you have a surrogate key and whether or not you have the luxury of taking a table down. Approaches vary from using subselects, dropping a table and rebuilding using a distinct query from temp table, and using non-set based approaches such as cursors.
The approach we often use is this one:
Continue reading "Deleting Duplicate Records in a Table"
Wednesday, January 09. 2008
Question: What is the answer to SELECT 3/2?
Answer: In integer math, it is 1. A lot of people especially those coming from MySQL or MS Access backgrounds are surprised to find out that in PostgreSQL
In actuality, the fact that 3/2 = 1 and 1/3 = 0 is part of the ANSI/ISO-SQL standard that states mathematical operations between two values must be of the same data type of one of the values (not necessarily the same scale and precision though). This is not some idiosyncracy specific to PostgreSQL. If you try the same operation in SQL Server, SQLite,FireBird, and some other ANSI/ISO SQL compliant databases, you will get the same results. So it seems MySQL and MS Access are the odd-balls in this arena, but arguably more practical.
Continue reading "SQL Math Idiosyncracies"
Saturday, January 05. 2008
Below is a Thumbnail view of a PostgreSQL 8.3 Cheat Sheet that covers prior PostgreSQL constructs plus new 8.3 features. PDF version of this cheat sheet is available at PostgreSQL 8.3 Cheat sheet in PDF 8/12 by 11", PostgreSQL 8.3 Cheat sheet in PDF A4 and the PostgreSQL 8.3 Cheat sheet in HTML.
We apologize for the small size of the fonts. We tried to keep it to a minimum but had trouble deciding what to leave out to keep it to a one page sheet. So perhaps it would have been better as a 2 part cheatsheet. Anyrate we hope people find it useful.
For people who love cheat sheets, here are two other PostgreSQL cheat sheets on the web which we have found useful and cover slightly different set of things than we do in ours.
Wednesday, January 02. 2008
Welcome to the January 2008 Issue of Postgres OnLine Journal Magazine. In this issue we will have a special feature PostgreSQL 8.3 Cheatsheet to commemorate the upcoming PostgreSQL 8.3 release and the new year. This cheat sheet will look similar in format to the Postgis Cheatsheet and will cover standard PostgreSQL features as well as new features added to the 8.3 release.
In future issues we hope to provide similar cheatsheets that highlight certain PostgreSQL advanced and specialty features. Any thoughts on what topics people would like to see in a cheatsheet are welcome.
Other interesting topics that will be covered in this issue to name a few
On another note - check out Andrew Dunstan's, minimum update Trigger. It will be nice to see this make it into the PostgreSQL 8.4 release. Granted we haven't had much of a need of this feature, but when you need it, it comes in very handy as demonstrated in Hubert Lubaczewski's related article Avoiding Empty Updates. We remember the first time we started working on MySQL a long long time ago - MySQL had this built in, but you couldn't turn it off. In certain situations such as when you have triggers this feature is often a misfeature. Granted I guess there are only a few cases where having this automatically on could be annoying especially when all the other Databases you work with don't do this and there is probably some overhead involved with checking which may not always outweigh the update/logging cost. Any rate as far as check-off lists goes for people who consider this a feature, it will be nice to cross this off the list as one reason why one would choose MySQL over PostgreSQL and better yet in PostgreSQL it is optional.
Monday, December 31. 2007
Have you ever noticed that in PostgreSQL you can put set returning functions in the SELECT part of an sql statement if the function is written in language SQL or C. Try the same trick for PL written functions such as plpgsql, plperl, plr etc, and you get a slap on the wrist of the form ERROR: set-valued function called in context that cannot accept a set. For Plpgsql and other PL languages you must put the set returning function in the FROM clause.
Below is a simple example:
So it appears that PostgreSQL is not quite as democratic as we would like.
All interesting, but so what? you may ask. It is bad practice to put set returning functions in a SELECT clause. Such things are commonly mistakes and should be avoided.
Functional Row Expansion
It turns out that there are a whole class of problems in SQL where the simplest way to achieve the desired result is via a technique we shall call Functional Row Expansion. By that, we mean that for each record in a given set, we want to return another set of records that can not be expressed as a constant join expression. Basically the join expression is different for each record or the function we want to apply is too complicated to be expressed as a static join statement or join at all.
Taking the above example. Lets say for each record in test, you want to return the 4 records preceding including the current one. So basically you want to explode each row into 5 or fewer rows. Your general gut reaction would be do something as follows:
these give error: ERROR: function expression in FROM cannot refer to other relations of same query level
--But this does what you want
Keep in mind what makes the above tricky is that you want to return at most 4 of the preceding plus current. If you want to return all the preceding plus current, then you can do a trivial self join as follows:
So as you can see - its sometimes tricky to tell when you need to use this technique and when you don't.
For this trivial example, writing the function as an SQL only function works fine and is the best to use. SQL functions unfortunately lack the ability to define dynamic sql statements, among other deficiencies so resorting to using a pl language is often easier which means you lose this useful feature of sql functions. Stuffing a pl function in an SQL function just might do the trick. We haven't tried this on other pl languages except plpgsql, but we suspect it should work the same.
Friday, December 28. 2007
Recommended Books: Fundamentals of Database Design
Question: Does PostgreSQL support stored procedures?
Short Answer: Sort Of as Stored functions.
For all intents and purposes, PostgreSQL has less of a need for CREATE PROCEDURE than other databases aside from looking more like other databases. For example in SQL Server -> 2005 - although you can write functions that return tables and so forth, you have to resort to writing CLR functions marked as unsafe to actually update data in a stored function. This gets pretty messy and has its own limitations so you have no choice but to use a stored procedures, which can not be called from within an SQL query. In MySQL 5.1 the abilities of functions are even more limiting - they can't even return a dataset. In PostgreSQL, you can write a function marked as VOLATILE that updates data and that can do all sorts of wacky things that are useful but considered by some to be perverse such as the following:
Another thing stored procedures can usually do that functions can not is to return multiple result sets. PostgreSQL can simulate such behavior by creating a function that returns a set of refcursors. See this .NET example Getting full results in a DataSet object: Using refcursors way down the page, that demonstrates creating a postgresql function that returns a set of refcursors to return multiple result sets using the Npgsql driver.
Prior to PostgreSQL 8.1, people could yell and scream, but PostgreSQL doesn't support Output Parameters. As weird as it is for a function to support such a thing, PostgreSQL 8.1+ do support output parameters and ODBC drivers and such can even use the standard CALL interface to grab those values.
At a glance it appears that PostgreSQL functions do all that stored procedures do plus more. So the question is, is there any reason for PostgreSQL to support bonafide stored procedures aside from the obvious To be more compatible with other databases and not have to answer the philosophical question, But you really don't support stored procedures?.
There must be some efficiency benefits to declaring something as a store procedure and having it called in that way. Not quite sure if anyone has done benchmarks on that. So for the time being PostgreSQL functions have the uncanny role of having a beak like a duck and the flexibility of a beaver, but having the makeup of a Platypus.
Thursday, December 27. 2007
Recommended Books: PostgreSQL 8.4 Internals and Appendixes (contribs) SQL Queries for Mere Mortals SQL Visual Quick Start
The generic way of doing cross tabs (sometimes called PIVOT queries) in an ANSI-SQL database such as PostgreSQL is to use CASE statements which we have documented in the article What is a crosstab query and how do you create one using a relational database?.
In this particular issue, we will introduce creating crosstab queries using PostgreSQL tablefunc contrib.
Tablefunc is a contrib that comes packaged with all PostgreSQL installations - we believe from versions 7.4.1 up (possibly earlier). We will be assuming the one that comes with 8.2 for this exercise. Note in prior versions, tablefunc was not documented in the standard postgresql docs, but the new 8.3 seems to have it documented at http://www.postgresql.org/docs/8.3/static/tablefunc.html.
Often when you create crosstab queries, you do it in conjunction with GROUP BY and so forth. While the astute reader may conclude this from the docs, none of the examples in the docs specifically demonstrate that and the more useful example of crosstab(source_sql,category_sql) is left till the end of the documentation.
To install tablefunc simply open up the share\contrib\tablefunc.sql in pgadmin and run the sql file. Keep in mind that the functions are installed by default in the public schema.
If you want to install in a different schema - change the first line that reads
Alternatively you can use psql to install tablefunc using something like the following command:
We will be covering the following functions
There are a couple of key points to keep in mind which apply to both crosstab functions.
Setting up our test data
For our test data, we will be using our familiar inventory, inventory flow example. Code to generate structure and test data is shown below.
Using crosstab(source_sql, category_sql)
For this example we want to show the monthly usage of each inventory item for the year 2007 regardless of project. The crosstab we wish to achieve would have columns as follows: item_name, jan, feb, mar, apr, may, jun, jul, aug, sep, oct, nov, dec
--Resulting crosstab query --Note: For this we don't need the order by month since the order of the columns is determined by the category_sql row order
The output of the above crosstab looks as follows:
crosstab(source_sql) is much trickier to understand and use than the crosstab(source_sql, category_sql) variant, but in certain situations and certain cases is faster and just as effective. The reason why is that crosstab(source_sql) is not guaranteed to put same named buckets in the same columns especially for sparsely populated data. For example - lets say you have data for CSCL for Jan Mar Apr and data for Phenol for Apr. Then Phenols Apr bucket will be in the same column as CSCL Jan's bucket. This in most cases is not terribly useful and is confusing.
To skirt around this inconvenience one can write an SQL statement that guarantees you have a row for each permutation of Item, Month by doing a cross join. Below is the above written so item month usage fall in the appropriate buckets.
In actuality the above query if you have an index on action_date is probably more efficient for larger datasets than the crosstab(source, category) example since it utilizes a date range condition for each month match.
There are a couple of situations that come to mind where the standard behavior of crosstab of not putting like items in same column is useful. One example is when its not necessary to distiguish bucket names, but order of cell buckets is important such as when doing column rank reports. For example if you wanted to know for each item, which projects has it been used most in and you want the column order of projects to be based on highest usage. You would have simple labels like item_name, project_rank_1, project_rank_2, project_rank_3 and the actual project names would be displayed in project_rank_1, project_rank_2, project_rank_3 columns.
Output of the above looks like:
Tricking crosstab to give you more than one row header column
Recall we said that crosstab requires exactly 3 columns output in the sql source statement. No more and No less. So what do you do when you want your month crosstab by Item, Project, and months columns. One approach is to stuff more than one Item in the item slot by either using a delimeter or using an Array. We shall show the array approach below.
Result of the above looks as follows:
Building your own custom crosstab function
If month tabulations are something you do often, you will quickly become tired of writing out all the months. One way to get around this inconvenience - is to define a type and crosstab alias that returns the well-defined type something like below:
Then you can write the above query as
Adding a Total column to the crosstab query
Adding a total column to a crosstab query using crosstab function is a bit tricky. Recall we said the source sql should have exactly
3 columns (row header, bucket, bucketvalue). Well that wasn't entirely accurate. The crosstab(source_sql, category_sql) variant of the function
allows for a source that has columns row_header, extraneous columns, bucket, bucketvalue.
Don't get extraneous columns confused with row headers. They are not the same and if you try to use it as we did for creating multi row columns, you will
be leaving out data. For simplicity here is a fast rule to remember.
Resulting output of our cross tabulation with total column looks like this:
If per chance you wanted to have a total row as well you could do it with a union query in your source sql. Unfotunately PostgreSQL does not support windowing functions that would make the row total not require a union. We'll leave that one as an exercise to figure out.
Another not so obvious observation. You can define a type that say returns 20 bucket columns, but your actual crosstab need not return up to 20 buckets. It can return less and whatever buckets that are not specified will be left blank. With that in mind, you can create a generic type that returns generic names and then in your application code - set the heading based on the category source. Also if you have fewer buckets in your type definition than what is returned, the right most buckets are just left off. This allows you to do things like list the top 5 colors of a garment etc.
Tuesday, December 18. 2007
Recommended Books: Getting Started with OpenOffice.org 3 written by OO.org group Database Programming with OO Base
For those who are not familiar with OpenOffice Base. OpenOffice Base is the equivalent of Microsoft Access in the OpenOffice Open source suite. While it is not as feature rich as Microsoft Access, it has been getting increasingly better and has some unique features that even Microsoft Access lacks. Unfortuantely you can't just convert an access mdb to its format like you can with other Open office suite products - Word to Writer Writer to Word etc. However you can open MS Access databases in OOBase, but you can't take advantage of the forms and reports in an MS Access Database.
One thing I always liked about Microsoft Access was the ease with which you could link to various different kinds of datasources and generate rapid queries and so forth. Microsoft Access has a particular feature called Access Projects which ties it very closely with Microsoft SQL Server. What an MS Access Project does is connect you with a specific SQL Server database and allow you to browse all the objects, create forms and reports etc against the objects etc. Unfortunately MS Access Project only works with SQL Server. For other datasources you need to use linked tables and can't make design changes and browse a database as you can with Access Projects.
We had looked at Openoffice Base a while ago and thought they are making progress, but still not quite good enough to put to daily use. When we revisited Open Office Base recently, we were surprised to find a couple of neat nuggets.
In the next couple of sections we'll lay out how to quickly setup OpenOffice, install the native PostgreSQL driver and JDBC PostgreSQL driver and connect to a PostgreSQL database in OpenOffice Base. Please forgive us for using Windows paths in this. We figured it would be easier for people to follow since most users coming to this site are windows users and a lot of Linux users already use OO and paths are too different from Linux/Mac OSX installs.
Continue reading "Using OpenOffice Base 2.3.1 with PostgreSQL"
Saturday, December 15. 2007
Recommended Books: PostGIS In Action PostgreSQL 8.4 Official The SQL Language PostgreSQL 8.4 Server Administration
In later issues we'll be covering other PostgreSQL contribs. We would like to start our first issue with introducing, PostGIS, one of our favorite PostgreSQL contribs. PostGIS spatially enables PostgreSQL in an OpenGeospatial Consortium (OGC) compliant way. PostGIS was one reason we started using PostgreSQL way back in 2001 when Refractions released the first version of PostGIS with the objective of providing affordable basic OGC Compliant spatial functionality to rival the very expensive commercial offerings. There is perhaps nothing more powerful in the geospatial world than the succinct expressiveness of SQL married with spatial operators and functions. Together they allow you to manipulate and analyze space with a single sentence. For details on using Postgis and why you would want to, check out the following links
Just as PostgreSQL has grown over the years, so too has PostGIS and the whole FOSS4G ecosystem. PostGIS has benefited from both the FOSS4G and PostgreSQL growths. On the PostgreSQL, improvements such as improved GIST indexing, bitmap indexes etc and on the FOSS4G side dependency projects such as Geos and Proj4, and JTS, as well as more tools and applications being built on top of it.
In 2001 only UMN Mapserver was available to display PostGIS spatial data. As time has passed, UMN Mapserver has grown, and other Mapping software both Commercial and Open Source have come on board that can utilize PostGIS spatial data directly. On the FOSS side there are many, some being UMN Mapserver, GRASS, uDig, QGIS, GDAL/OGR, FeatureServer, GeoServer, SharpMap, ZigGIS for ArcGIS integration, and on the commercial side you have CadCorp SIS, Manifold, MapDotNet, Safe FME Data Interoperability and ETL tools.
In terms of spatial databases, PostGIS is the most capable open source spatial database extender. While MySQL does have some spatial capabilities, its spatial capabilities are extremely limited particularly in the selectivity of the spatial relational functions which are all MBR only, ability to create spatial indexes on non-MyISAM stores, and lack a lot of the OGC compliant functions such as Intersection, Buffering even in its 5.1 product. For details on this check the MySQL 5.1 docs - Spatial Extensions.
When compared with commercial spatial databases, PostGIS has most of the core functions you will see in the commercial databases such as Oracle Spatial, DB2 Spatial Blade, Informix Spatial Blade, has comparable speed, fewer deployment headaches, but lacks some of the advanced add-ons you will find, such as Oracle Spatial network topology model, Raster Support and Geodetic support. Often times the advanced spatial features are add-ons on top of the standard price of the database software.
Some will argue that for example Oracle provides Locator free of charge in their standard and XE versions, Oracle Locator has a limited set of spatial functions. Oracle's Locator is missing most of the core spatial analysis and geometric manipulation functions like centroid, buffering, intersection and spatial aggregate functions; granted it does sport geodetic functionality that PostGIS is currently lacking. To use those non-locator features requires Oracle Spatial and Oracle Enterprise which would cost upwards of $60,000 per processor. Many have heard of SQL Server 2008 coming out and the new spatial features it will sport which will be available in both the express and the full version. One feature that SQL Server 2008 will have that PostGIS currently lacks is Geodetic support (the round world model so to speak). Aside from that SQL Server 2008 has a glarying omission from a current GIS perspective - and that is the ability to transform from one spatial reference system to another directly in the database and is Windows bound so not an option for anyone who needs or is thinking of cross-platform or in a Unix environment. SQL Server 2008 will probably come closest to PostGIS in terms of price / functionality. The express versions of the commercial offerings have many limitations in terms of size of database and usually limited to one processor use. For any reasonably sized deployment in terms of database size, processor utilization, replication, or ISP/Service Provider/Integrator this is not adequate and for any reasonably large deployment that is not receiving manna from heaven, some of the commercial offerings like Oracle Spatial, are not cost-sensible.
Note that in near future versions PostGIS is planning to have geodetic support and does provide basic network topology support via the PgRouting project and there are plans to incorporate network topology as part of PostGIS.
There is a rise in the use of mapping and geospatial analysis in the world and it is moving out of its GIS comfort zone to mingle more with other IT Infrastructure, General Sciences, and Engineering. Mapping and the whole Geospatial industry is not just a tool for GIS specialists anymore. A lot of this rise is driven by the rise of mapping mashups - things like Google Maps, Microsoft Virtual Earth, and Open data initiatives that are introducing new avenues of map sharing and spatial awareness. This new rise is what many refer to as NeoGeography. NeoGeography is still in its infancy; people are just getting over the excitement of seeing dots in their hometown, and are quickly moving into the next level - where more detailed questions are being asked about those dots and dots are no longer sufficient. We want to draw trails such as trail of hurricane destruction, avian bird flu, track our movement with GPS, draw boundaries and measure the densities of these based on some socio-ecological factor and we need to store all that user generated or tool generated information, and have all that transactional goodness, security and ability to query in an easy way that a relational database offers. This is the level where PostGIS and other spatial databases are most useful.
Monday, December 10. 2007
Recommended Books: Blogging for Dummies Smarty PHP Template Programming And Applications Blogging Heroes - Interview with top 30 bloggers
Choosing Blogging Software
When we started blogging, we had several criteria for the blogging software we would use.
We immediately dismissed wordpress because it was MySQL centric, Blogger etc services were out the door as well. There were not that many blogging applications in .NET and most were very SQL Server centric.
We noticed other PostgreSQL bloggers use predominantly Serendipity, so we thought we'd give it a try.
Serendipity met all our requirements except for the PHP ADODB part. It has a database abstraction layer, but it appears to be a custom one. This we could live with. Below are the features we really liked about it.
There are some plugins enabled by default, but can't remember which ones. For the most part they are the common ones people would choose if they chose them. These get you pretty far at least to use the software before you realize hey there is other stuff you can turn on or off. Below are some of the ones we found as must haves or things that should think about changing.
Serendipity has plugins broken up into event plugins and side bar plugins. Side bar plugins can be drag and dropped between the left right middle areas, which is a nice convenience. Event plugins are triggered based on Serendipity system events such as blog posts or comment posts and some aren't really events so to speak but aren't side bar plugins either so they show under events.
We haven't played with these too much. The standard default calendar, category, and search were pretty much what we needed starting off. We liked the Wiki Finder and the links to publish to social bookmarking sites as a nice convenience.
We also tried this on a virgin install of PostgreSQL 8.3 Beta 3 and it didn't work. Seems to be some logic in the DB layer of serendipity that uses LIKE instead of = against ids and the fact that PostgreSQL 8.3 has taken out a lot of the default CASTS. I think the serendipity code should be changed in this case since from a cursory glance, doesn't quite look right or efficient, but I'm sure there is a good reason they chose to do things that way.
Friday, December 07. 2007
One of the annoying things about PostgreSQL unlike some other databases we have worked with is that simple views are not automatically updateable. There is some work involved to make views updateable. For simple views, this is annoying, but for more complex views it is a benefit to be able to control how things are updated. In a later version of PostgreSQL perhaps 8.4 or 8.5 this will be ratified and PostgreSQL will enjoy the same simplicity of creating simple updateable views currently offered by MySQL and SQL Server and other DBMSs, but still allow for defining how things should be updated for more complex views. For this exercise we are using PostgreSQL 8.2.5, but most of it should work for lower versions with slight modification.
For this exercise, we shall create a fairly complex updateable view to demonstrate how one goes about doing this.
Here is a scenario where being able to control how a view is updated comes in very handy.
We all know relational databases are great because they give you great mobility on how you slice and dice information. At times for data entry purposes, the good old simple flat file is just more user-friendly.
Problem: You are developing an inventory application for a molecular biology lab and they have the following requirements:
They have 2 projects going on. One on Multiple Sclerosis Research (MS) and one on Alzheimer's. Each is funded by different grants and for grant cost allocation purposes, they need to keep track of the supplies they use on each project.
In our system we have 2 tables for simplicity. inventory and inventory_flow. I know we should have a project lookup table or in 8.3 possibly use an ENUM, but to make this short, we are skipping that.
Now look at what happens when we insert and update our view
The slick thing about this is if you were to create a linked table in something like say Microsoft Access and designated item_id as the primary key, then the user could simply open up the table and update as normally and behind the scenes the rules would be working to do the right thing.
Thursday, December 06. 2007
PostgreSQL 8.3 is currently in Beta 4 and promises to offer some whoppingly neat features. First before we go over the new features we are excited about in this upcoming release, we'd like to briefly cover what was added in past releases.
The big 8.0 Highlights
8.3 upcoming Highlights
8.3 has numerous highlights just as previous versions, but we shall focus on our favorite ones.
Wednesday, December 05. 2007
One of the features in PostgreSQL designed to enhance index performance is the use of a clustered index. For people coming from MS SQL Server shops, this may look familiar to you and actually serves the same purpose, but is implemented differently and this implementation distinction is very important to understand and be aware of. In PostgreSQL 8.3 the preferred syntax of how you cluster has changed. For details check out 8.3 CLUSTER 8.2 CLUSTER 8.0 CLUSTER. A lot of what I'm going to say is somewhat of a regurgitation of the docs, but in slightly different words.
First in short - clustering on an index forces the physical ordering of the data to be the same as the index order of the index chosen. Since you can have only one physical order of a table, you can have only one clustered index per table and should carefully pick which index you will use to cluster on or if you even want to cluster. Unlike Microsoft SQL Server, clustering on an index in PostgreSQL does not maintain that order. You have to reapply the CLUSTER process to maintain the order. Clustering helps by reducing page seeks. Once an index search is done and found, pulling out the data on the same page is vastly faster since once you find the start point all successive data nearby is easy picking.
As a corrollary to the above, it doesn't help too much for non-range queries. E.g. if you have dummy ids for records and you are just doing single record select queries, clustering is fairly useless to you. It is only really useful if you are doing range queries like between date ranges or spatial ranges or queries where the neighboring data to an index match is likely to be pulled. For example if you have an order items table, then clustering on a compound index such as order_id,order_item_id may prove useful since neighboring data is something you likely want to pull for range and summations.
Now lets see how we create a clustered index and then talk about the pros and gotchas
Once a clustered index is created to force a recluster, you simply do this
To force a cluster on all tables that have clustered indexes, you do this
What is FillFactor and how does it affect clustering?
Again those coming from Microsoft SQL Server will recognize FILLFACTOR syntax. IBM Informix also has a FILLFACTOR syntax that serves the same purpose as the SQL Server and PostgreSQL ones. For more details here PostgreSQL docs: Create Index. FillFactor basically creates page gaps in an index page. So a Fill Factor of 80 means leave 20% of an index page empty for updates and inserts to the index so minimal reshuffering of existing data needs to happen as new records are added or indexed fields are updated in the system. This is incorporated into the index creation statement.
After an index is created on a table, this information is then used in several scenarios
Why should you care?
First for fairly static tables such as large lookup tables, that rarely change or when they change are bulk changes, there is little point in leaving blank space in pages. It takes up disk space and causes Postgres to scan thru useless air. In these cases - you basically want to set your FillFactor high to like 99.
Then there are issues of how data is inserted, if you have only one index and new data usually resides at the end of the index and the indexed field are rarely updated, again having a low fill factor is probably not terribly useful even if the data is updated often. You'll never be using that free space so why have it.
For fairly updated data that changes such that you are randomly adding 10% new data per week or so in middle of page, then a fill factor of say 90 is the general rule of thumb.
Cluster approach benefits and Gotchas
The approach PostgreSQL has taken to cluster means that unlike the SQL Server approach, there is no additional penalty during transactions of having a clustered index. It is simply used to physically order the data and all new data goes to the end of the table. In the SQL Server approach all non-clustered indexes are keyed by the clustered index, which means any change to a clustered field requires rebuilding of the index records for the other indexes and also any insert or update may require some amount of physical shuffling. There are also other consequences with how the planner uses this information that are too detailed to get into.
The Bad and the Ugly
The bad is that since there is no additional overhead aside from the usual index key creation during table inserts and updates, you need to schedule reclustering to maintain your fine order and the clustering causes a table lock. The annoying locking hopefully will be improved in later versions. Scheduling a cluster can be done with a Cron Job or the more OS agnostic PgAgent approach. In another issue, we'll cover how to use PgAgent for backup and other scheduling maintenance tasks such as this.