Related
What are the reasons behind Microsoft implementing its own SQL extension as Transact SQL (T-SQL)? What are its advantages over the normal SQL?
Everybody extends SQL.
SQL isn't procedural, it's declarative. You describe what you want, and it figures out how to retrieve it using whatever indexes or hashes or whatnot is available.
But sometimes, that's not sufficient. T-SQL gives syntax to run procedural code inside of your queries. This lets you do control structures (begin-end, if-then-else), iteration and move values between local variables, temporary tables and other sources.
Transact-SQL (T-SQL) is Microsoft's and Sybase's proprietary extension to SQL. Microsoft's implementation ships in the Microsoft SQL Server product. Sybase uses the language in its Adaptive Server Enterprise, the successor to Sybase SQL Server.
Transact-SQL enhances SQL with these additional features:
Control-of-flow language
Local variables
Various support functions for string processing, date processing, mathematics, etc.
Improvements[citation needed] to DELETE and UPDATE statements
http://en.wikipedia.org/wiki/Transact-SQL
Wiki connects you to much more expanding information and details.
What are the reasons behind Microsoft implementing its own SQL extension as Transact SQL (T-SQL)?
To make your life easier.
What are its advantages over the normal SQL?
There is no such thing as "normal SQL"
Transact-SQL both enhances the set-based abilities of SQL and adds procedural abilities.
Other systems (like Oracle and PostgreSQL) clearly distinguish between SQL and procedural languages (PL/SQL and pl/PgSQL).
Microsoft doesn't make such a strict distinction.
Transact-SQL was developed by Sybase around the middle of 80's, when there was no standard at all (the first one was proposed in 1986).
By that time each vendor already had a burden of legacy applications to support, and rewriting their databases to conform to the standard would break the compatibility.
There is more or less commonly supported standard, SQL:92, but it still misses very very much to be really of use.
That's why almost every task beyond a simple SELECT with a JOIN needs some proprietary support to be implemented efficiently.
A thing to note is that while most RDBMS providers make a clear distinction between their extensions to SQL and the programming languages used to write stored procedures and triggers and so on, Microsoft and Sybase do exactly the contrary, they mix these two concepts into one, namely, T-SQL. You use T-SQL when you write normal queries, but you also can (and usually do) use T-SQL when you are writing stored procedures and triggers.
This has the controversial benefit of encouraging (or at least making very easy) the creation a mix of procedural and SQL code[*].
Nowadays Microsoft makes a distinction between T-SQL stored procedures and those written for the CLR (ie, .NET), but this is a relatively new development (from SQL Server 2005 onwards).
[*]: Controversial because people who don't speak SQL will be tempted to write procedural code (usually very inefficient in databases) instead of learning SQL (the proper thing to do).
The applicable SQL Standard for control-of-flow, local variables, etc (i.e. procedural code) is known as SQL/PSM (Persistent Stored Modules).
According to Wikipedia it was adopted in 1996 but I suspect it's the usual problem: vendors were already committed to their own extensions and therefore take up of the Standards is postponed for long periods of time
...but not necessarily indefinitely, there is hope. For example, common table expressions (CTEs) and OLAP functions in SQL Server 2005 and temporal data types in SQL Server 2008 indicate that extensions to TSQL will keep close to the publish Standards.
One other really important reason why vendors create their own flavors of SQL is performance tuning. There are many ways of writing more performant queries using vendor specific code that is written to take advantage of how that particular database engine works.
I'm developing an asp.net application with Database factory pattern which allows the application to support both Sql Server and Oracle. I've created an abstract class that has the methods common to Sql Server and Oracle, like the CreateConnection and CreateCommand methods. This class is implemented by SqlServer and Oracle classes. Now, is there an easy way to write in-line sql queries with parameters common to both Sql Server and Oracle. I mean, I understand that we use "#" symbol in Sql Server and ":" in Oracle for parameters. Just for this reason, I'm writing queries twice in each of the class. Is there a way to write such queries common to both the databases? (or interpret the parameters from one common query?)
Thanks.
The only way to write one query that will work for both Oracle and Sql Server is to use only the syntax that is common to both platforms. Once you use features that are different between the two languages (like parameters or joins), you either have to write two different queries or hack together a "translator" class that converts a query from one platform to the other.
I've done a lot of this type of programming (database-agnostic software), and with .Net a relatively pain-free way of doing this is to write your main application to work entirely with ADO.Net DataTables/DataSets, with a wrapper class that handles generating the DataTables from either Oracle or Sql Server tables under-the-hood, and also handles persisting changes made to the DataTables back into Oracle or Sql Server. This approach isolates your DB-specific code in one place, although it's not necessarily a viable approach if the data your application needs access to is large.
You could write some kind of translator, but I would suggest that in some cases you'll need to write db-specific code for performance reasons anyway, so you'll have to put up with the maintenance burden of two versions of some queries.
What is the point of using ORACLE and not using all its non standard functions (analytics, pivots etc) ? ORACLE is a powerful tool.
Other DBs have there own strenght also, so why use the lowest common denominator just to be able to work on ALL of them? You will just lose in performance.
Just pick one DB, and use it fully with all its functionalities !
Pardon my ignorance here, but can't something like an ORM (object relational mapper) work for both SQL and Oracle?
I had similar requirements, to support both Sql Server and Oracle, and summarized my two years of experience with such problems in these articles:
Writing ANSI Standard SQL is not practical.
Think ANSI Standard SQL Is Fully Portable Between Databases? Think Again.
The application I want to build using MS Visual C# Express (I'm willing to upgrade to Standard if that becomes required) that needs a database.
I was all psyched about the SQL Server Compact - because I don't want the folks who would be installing my application on their computers to have to install the whole of SQL Server or something like that. I want this to be as easy as possible for the end user to install.
So I was all psyched until it seems that there are limitations to what I can do with the columns in my tables. I created a new database, created a table and when I went to create columns it seems that there isn't a "text" datatype - just something called "ntext" that seems to be limited to 255 characters. "int" seems to be limited to 4 (I wanted 11). And there doesn't seem to be an "auto_increment" feature.
Are these the real limitations I would have to live with? (Or is it because I'm using "Express" and not "Standard"). If these are the real limitations, what are my other database options that meet my requirements? (easy installation for user being the biggie - I'm assuming that my end user is just an average user of computers and if it's complicated would get frustrated with my application)
-Adeena
PS: I also want my database data to be encrypted to the end user. I don't want them to be able to access the database tables directly.
PPS. I did read: http://www.microsoft.com/Sqlserver/2005/en/us/compact.aspx and didn't see a discussion on these particular limitations
I'm not sure about encryption, but you'll probably find this link helpful:
http://msdn.microsoft.com/en-us/library/ms171955.aspx
As for the rest of it:
"Text" and "auto_increment" remind me of Access. SQL Server Compact is supposed to be upgrade compatible to the server editions of SQL Server, in that queries and tables used in your compact database should transfer to a full database without modification. With that in mind, you should first look at the SQL Server types and names rather than Access names: in this case namely varchar(max), bigint, and identity columns.
Unfortunately, you'll notice this fails with respect to varchar(max), because Compact Edition doesn't yet have the varchar(max) type. Hopefully they'll fix that soon. However, the ntext type you were looking at supports many more than 255 bytes: 230 in fact, which amounts to more than 500 million characters.
Finally, bigint uses 8 bytes for storage. You asked for 11. However, I think you may be confused here that the storage size indicates the number of decimal digits available. This is definitely NOT the case. 8 bytes of storage allows for values up to 264, which will accomodate many more than 11 digits. If you have that many items you probably want a server-class database anyway. If you really want to think in terms of digits, there is a numeric type provided as well.
A few, hopefully helpful comments:
1st - do not use SQLite unless you like having to have the entire database locked during writes (http://www.sqlite.org/faq.html#q6) and perhaps more importantly in a .Net application it is NOT thread safe or more to the point it must be recompiled to support threads (http://www.sqlite.org/faq.html#q6)
As an alternate for my current project I looked at Scimore DB (they have an embedded version with ADO.Net provider: http://www.scimore.com/products/embedded.aspx) but I needed to use LINQ To SQL as an O/RM so I had to use Sql Server CE.
The auto increment (if you are referring to automatic key incrementing) is what it always has been - example table:
-- Table Users
CREATE TABLE Tests (
Id **int IDENTITY(1,1) PRIMARY KEY NOT NULL,**
TestName nvarchar(100) NOT NULL,
TimeStamp datetime NOT NULL
)
GO
As far as the text size I think that was answered.
Here is a link to information on encryption from microsoft technet: (http://technet.microsoft.com/en-us/library/ms171955.aspx)
Hope this helps a bit....
Had to chime in on two factors:
I use Sql Compact a lot and its great for what it works for -- a single user, embedded, database, with a single file data store. It has all the SQL goodness and transactions. It hadles parallellism well enough for me. Notice that few of the naysayers on this page use the product regularly. Don't use it on a server, that's not what its for. Many of my customers don't even know the file is a "database", that is just an implementation issue.
You want to encrypt the data from your users -- presumably so they can only view it from your program. This simply isn't going to happen. If your program can decrypt the data, then you have to store the key somewhere, and a sufficently dedicated attacker will find it, period.
You may be able to hide the key well enough that the effort to recover it isn't worth the value of the information. Windows has some neat machine and user local encryption routines to help. But if your design has a strong requirement that a user never find data you have hidden on their computer (but your program will) you need to redesign -- that guarentee simply cannot be accomplished.
SQL CE is a puzzle to me. Did we really need yet another different SQL database platform? And it's the third in the last several years targeted at mobile platforms from MS ... I wouldn't have a lot of confidence that it will be the final one. It doesn't share much if any technology with SQL Server - it's a new one from scratch as far as I can tell.
I've tried it, and then been more successful with both SQLite and Codebase.
EDIT: Here is a list of the (many) differences.
ntext supports very large text data (see MSDN - this is for Compact 4.0, but the same applies to 3.5 for the data types you are mentioning).
int is a numeric data type, so the size of 4 means 4 bytes/32 bits of storage (–2,147,483,648 to 2,147,483,647). If you intend to store 11 bytes of data in a single column, use the varbinary type with a size of 11.
Automatically incrementing columns in the SQL Server world are done using the IDENTITY keyword. This causes the value of the column to be automatically determined by SQL Server when inserting data into a row, preventing collisions with any other rows.
You can also set a password or encrypt the database when creating it in SQL Compact to prevent users from directly accessing your application. See Securing Databases on MSDN.
All of the items you mention above are not really limitations, so much as they are understanding how to use SQL Server.
Having said that, there are some limitations to SQL Compact.
No support for NVARCHAR(MAX)
NTEXT works just fine for this
No support for VIEWs or PROCEDUREs
This is what I see as the primary limitation
I've used the various SQL Server Compact editions on a few occasions, but only ever as data capture repositories on mobile platforms - where it works well for syncing with a server database, and with that sort of scenario is undoubtedly the optional choice.
However if you need something to do more than that and act as a primary database to your application then I'd suggest SQLLite is probably the better option, it's completely solid, widely supported and found in all sorts of places (used on the iPhone for example) but is surprisingly capable (The Virtual Reality simulator OpenSim uses it as it's default database) and there are lots of others (including Microsoft).
I must also chime in here with VistaDB as an alternative to SQL CE.
VistaDB does support encryption (Blowfish), it also supports TEXT as well as NTEXT (including FTS indexes on them).
And yes the post above is correct in that you have to look at the SQL Server types to really match them up, VistaDB also uses the SQL Server types (we actually support more than SQL CE does; only missing XML).
To see other comparisons between VistaDB and SQL CE visit the comparison page. Also see the SO thread on Advantages of VistaDB for more information.
(Full disclosure - I am the owner of VistaDB so I may be biased)
According to this post (http://www.nelsonpires.com/web-development/microsoft-webmatrix-the-dawn-of-a-new-era/) it says that because it uses a database file, only one process can access it for every read/write and as a result it needs exclusive access to the file, also it is limited to 256 connections and the whole file will most likely have to be loaded in memory. So SQL server compact might not be good for your site when it grows.
There are constraints... Joel seems to have addressed the details. SQL CE is really geared for mobile development. Most of the "embedded" database solutions have similar constraints. Check out
SQLite
No TEXT field character limit
Auto increment only on INTEGER PRIMARY KEY column
Some third party encryption support
Esent
(unmanaged code isn't my forte, and I can't decipher the unmanaged docs)
Is it possible to make efficient queries that use the complete regular expression feature set.
If not Microsoft really should consider that feature.
For SQL Server 2000 (and any other 32 bit edition of SQL Server), there is xp_pcre, which introduces Perl compatible regular expressions as a set of extended stored procedures. I've used it, it works.
The more recent versions give you direct access to the .NET integrated regular expressions (this link seems to be dead, here is another one: MSDN: How to: Work with CLR Database Objects).
The answer is no, not in the general case, although it might depend on what you mean by efficient. For these purposes, I'll use the following definition: 'Makes effective use of indexes and joins in a sensible order' which is probably as good as any.
In this case, 'Efficient' queries are 's-arg'-able, which means that they can use index lookups to narrow down search predicates. Equalities (t-joins) and simple inequalities can do this. 'AND' predicates can also do this. After that, we get into table, index and range scanning - i.e. operations that have to do record-by-record (or index-keyby index-key) comparisons.
Sontek's answer describes a method of in-lining regexp functionality into a query, but the operations still have to do comparisons on a record by record basis. Wrapping it up in a function would allow a function-based index where the result of a calculation is materialised in the index (Oracle supports this and you can get equivalent functionality in SQL Server by using the sort of tricks discussed in this article). However, you could not do this for an arbitrary regexp.
In the general case, the semantics of a regular expression do not lend themselves to pruning match sets in the sort of way that an index does, so integrating rexegp support into the query optimiser is probably not possible.
Check out this and this. They are great posts on how to do it.
I would love to have the ability to natively call regular expressions in SQL Server for ad hoc queries and use in stored procedures. Our DBA's won't allow us to create CLR functions so I have been using LINQ Pad as a kind of poor man's query editor for the ad hoc stuff. It is especially useful when working with structured data such as JSON or XML that has been saved to the database.
And I agree that it seems like an oversight that there is no regular expression support, it seems like an obvious feature for a query language. Hopefully we will see it in a future version but people have been asking for it for a long time and it hasn't made it's way into the product yet.
The most frequent reason I have seen against it is that a poorly formed expression can cause catastrophic backtracking which in .NET will not abort and almost always requires the machine to be restarted. Maybe once they address that in the framework we will see it included in a future version of SQL Server.
I think we can see from the new types in SQL Server 2008 (hierarchyid, geo-spatial) that if Microsoft do add this it will come in the form of a SQL CLR Assembly
If you are able to install Assemblies into your database you could roll your own by creating a new Database\SQL Server project in Visual Studio - this will allow you to make a new Trigger / UDF / Stored Proc / Aggregate or UDT. You could import System.Text.RegularExpressions into the class and go from there.
Hope this helps
Closed. This question needs to be more focused. It is not currently accepting answers.
Want to improve this question? Update the question so it focuses on one problem only by editing this post.
Closed 3 years ago.
Improve this question
We all have our favourite database. If you look objectively at your chosen database, what drawbacks does it have and what could be improved?
The rules:
One reply per drawback with;
a short description of the limitation, followed by;
a more detailed description, an explanation of how it could be done better or an example of another technology that does not have the same limitation.
Do not diss any database that you haven't used extensively. It is easy to take potshots at other technologies but we want to learn form your experience, not your prejudice.
Oracle databases are quite expensive
Oracle does what it does well but the licensing costs are horrendous. That has been improved by the release of Oracle XE but the limitations of that mean that it is a growth constraint on you solution.
Database Microsoft SQL Server 2005
Defect Lack of "INSERT OR UPDATE"
Description
Often you need to either insert or update a record in a table, depending on whether the record is present or not. Not having an atomic operation to do so leads to unnecessary transactions.
This does not happen with MySQL or SQLServer 2008.
Database PostgreSQL
Defect No SQL Profiler
We asked the developers about this at a recent conference and I understand it's now something they're looking to implement.
I love the flexibility of sequences in Oracle as compared with other databases autoincrements, but the inability to set seq.nextval as a default for a pk column is somewhat annoying, and must be trivial to fix.
Database Microsoft SQL Server
Defect Huge licensing cost
Description
SQL Server has great features and it integrates very well with .NET development. The issue is that when you have to scale up from a shared database to a dedicated database, licensing costs are really high. This, in effect, leads to databases which should really run on a dedicated server, being hosted on shared servers with performance and security issues.
This does not happen with MySQL or PostgreSQL.
Database Microsoft SQL Server 2005
Defect Badly implemented UI
Description
SQL Server management studio does not offer a great user experience:
Tabbing behaviour is weird: you are always looking for the right tab
Keeps on crashing on 64-bit versions
Missing some features of preceding version, like overview of grants of stored procedures
This does not happen with version 2000.
Database MySQL
Defect Server will start up with damaged tables
Description
If MySQL has a damaged table - from either being killed during a write or some other failure - it will quite happily start up and allow the user to carry on as if the problem does not exist. Granted it will produce some error messages in the log, but from my experience this doesn't help when you're trying to figure out why an application is behaving oddly.
Most other databases will detect and repair the error on startup or simply refuse to start with any sort of corruption.
Database MySQL 5.0.x and above
Defect Ring replication errors lead to inconsistent data on different nodes
Description
The most serious problem in production we face at the moment is that in a MySQL ring the ring itself produces an error and stops replicating.
Building a ring (or Master-Master-replication) is possible since 5.x.x: You chain the databases in a "ring" so that the replicate data to each other. Every database-node gets all the changes from all other nodes.
We assume that the error lies behind autoincrement- failures. This is known from normal replication, too, but in the new version there are no sufficinet error messages in the error log. I highly recommend not to use this feature in MySQL as long as the problems here are not fixed.
Database Oracle
Defect Did not handle long datatype well for too long
Description
Oracle only had the long datatype until 9i (I believe) at which point it was deprecated in favor of the LOBs. There is a ton of code out there, however, which still has longs and all of the related restrictions. The biggest of which was that each table could only have one long column and it had to be at the end of the columns. See here for a more exhaustive list of restricitons on the long.
Database Oracle
Problem Temp table definitions are not private
Description Many databases (eg Postgres and Sybase) allow you to create temp tables on the fly, insert into them, add indexes if you want, then query from them. Oracle has temp tables, but the temp table definitions exist in a global name space. Therefore the temp table has to be created by a DBA, you need to synchronize between the table definition they used and your code, and if two pieces of code want similar (but not identical) table definitions, they need to use different names. These differences make temp tables far less convenient for developers.
Yes, I understand the benefits for the query optimizer of having global definitions. However for me the lack of convenience makes Oracle's temp tables virtually useless for me, while I use them very intensively in Postgres.
Database: Oracle
Problem: The names of tables, procedures, columns, etc cannot exceed 30 characters. This is infuriating.
Problem: It's slapdash JDBC compliance. For example, stored procedures do not return results sets in a JDBC-compliant way, but instead of a proprietary OUT parameter type. This means you can't use higher-level JDBC abstractions.
Database MySQL
Defect Foreign Keys supported only on some table types
Description
Enough said. It has obvious maintenance implications.
From the MySQL manual
Foreign keys definitions are subject to the following conditions:
Both tables must be InnoDB tables and they must not be TEMPORARY tables.
And here:
For storage engines other than InnoDB, MySQL Server parses the FOREIGN KEY syntax in CREATE TABLE statements, but does not use or store it.
This does not happen with any other major DB.
PostgreSQL doesn't have a good failover solution, but I understand they're working on it.
Database : Sql Compact Edition
Drawback : Stored procedures are not supported.
Regardless of this limitation, this DB has its' uses especially as a client cache for application that can be smart client or distributed to mobile platforms.
Database Oracle
Defect Granularity of grants on packages
Description
You can only grant permissions on packages and not on stored procedures inside packages. Or alternatively, you can grant permissions on single stored procedures but then you put them outside of packages. This requires you to know up front who will use which stored procedure and it is really hard to refactor.
This does not happen with SQL Server.
Database Microsoft SQL Server 2005
Defect Lack of array type parameters
Description
Useful in searches, a lot of times you need to pass a series of values to be matched against. In SQL 2005 you can do a workaround by using CLR inside SQLServer. Given the usefulness it would make more sense to have this feature out of the box.
This does not happen with SQL Server 2008 or Oracle.
Database Postgres
Defect No analytic queries
Description
Analytic queries, introduced by Oracle, are part of the SQL 2003 standard. Unfortunately Postgres hasn't implemented them yet.
Database : PostgreSQL
**Problem : ** is that connector for C# for example are not really up-to-date and crash with advanced feature.
Database: All
Drawback - Poor design by people who didn't think it was important to know what you were doing when you designed a datbase. Far more problmes caused in all databases by bad design than from any missing feature. So I suppose they are all missing the "read my mind and figure out the best solution without me having to think" feature.
Any SQL DBMS
Defect: Duplicate rows
One of the virtues of the relational model is that it represents everything without duplicate tuples, i.e. using relations, which have keys and no duplicates. Unfortunately SQL isn't built that way. This makes the database developer's life needlessly difficult. SQL developers have to deal with tables without keys and debug queries that return duplicate rows.