CREATE DATABASE statement

Description

Creates a database consisting of several operating system files.

Syntax

CREATE DATABASE db-name
… [ [ TRANSACTION ] { LOG ON [ log-file-name ]
      [ MIRROR mirror-file-name ] } ]
… [ CASE { RESPECT | IGNORE } ]
… [ PAGE SIZE page-size ]
… [ COLLATION collation-label[( collation-tailoring-string ) ] ]
… [ ENCRYPTED [ TABLE ] {algorithm-key-spec | OFF } ]
… { … [ BLANK PADDING ON ]
… [ JAVA { ON | OFF } ]
… [ JCONNECT { ON | OFF } ]
… [ IQ PATH iq-file-name ]
… [ IQ SIZE iq-file-size ]
… [ IQ PAGE SIZE iq-page-size ]
… [ BLOCK SIZE block-size ]
… [ IQ RESERVE sizeMB ]
… [ TEMPORARY RESERVE sizeMB ]
… [ MESSAGE PATH message-file-name ]
… [ TEMPORARY PATH temp-file-name ]
… [ TEMPORARY SIZE temp-db-size ]

Parameters

Examples

Example 1

The following Windows example creates an IQ database named mydb with its corresponding mydb.db, mydb.iq, mydb.iqtmp, and mydb.iqmsg files in the C:\s1\data directory:

CREATE DATABASE 'C:\\s1\\data\\mydb'
BLANK PADDING ON
IQ PATH 'C:\\s1\\data'
IQ SIZE 2000
IQ PAGE SIZE 65536

Example 2

The following UNIX command creates an IQ database with raw devices for IQ PATH and TEMPORARY PATH. The default IQ page size of 128KB applies.

CREATE DATABASE '/s1/data/bigdb'
IQ PATH '/dev/md/rdsk/bigdb'
MESSAGE PATH '/s1/data/bigdb.iqmsg'
TEMPORARY PATH '/dev/md/rdsk/bigtmp'

Example 3

The following Windows command creates an IQ database with a raw device for IQ PATH. Note the doubled backslashes in the raw device name (a Windows requirement):

CREATE DATABASE 'company'
IQ PATH '\\\\.\\E:'
JCONNECT OFF
IQ SIZE 40

Example 4

The following UNIX example creates a strongly encrypted IQ database using the AES encryption algorithm with the key “is!seCret.”

CREATE DATABASE 'marvin.db'
JAVA OFF
BLANK PADDING ON
CASE RESPECT
COLLATION 'ISO_BINENG'
IQ PATH '/filesystem/marvin.main1'
IQ SIZE 6400
IQ PAGE SIZE 262144
TEMPORARY PATH '/filesystem/marvin.temp1'
TEMPORARY SIZE 3200
ENCRYPTED ON KEY 'is!seCret' ALGORITHM 'AES'

Usage

Creates an IQ database with the supplied name and attributes. The IQ PATH clause is required for creating the IQ database. Otherwise, you create a standard SQL Anywhere database. If you omit the IQ PATH option, specifying any of the following options generates an error: IQ SIZE, IQ PAGE SIZE, BLOCK SIZE, MESSAGE PATH, TEMPORARY PATH, and TEMPORARY SIZE.

When Sybase IQ creates an IQ database, it automatically generates four database files to store different types of data that constitute an IQ database. Each file corresponds to a dbspace, the logical name by which Sybase IQ identifies database files. The files are:

• db-name.db is the file that holds the catalog dbspace, SYSTEM. It contains the system tables and stored procedures describing the database and any standard SQL Anywhere database objects you add. If you do not include the .db extension, Sybase IQ adds it. This initial dbspace contains the catalog store, and you can later add dbspaces to increase its size. It cannot be created on a raw partition.

• db-name.iq is the default name of the file that holds the main data dbspace, IQ_SYSTEM_MAIN, containing the IQ tables and indexes. You can specify a different file name with the IQ PATH clause. This initial dbspace contains the IQ store.

  IQ_SYSTEM_MAIN is a special dbspace that contains all structures necessary for the database to open: the IQ db_identity blocks, the IQ checkpoint log, the IQ rollforward/rollback bitmaps of each committed transaction and each active checkpointed transaction, the incremental backup bitmaps, and the freelist root pages. IQ_SYSTEM_MAIN is always online when the database is open.

  The administrator can allow user tables to be created in IQ_SYSTEM_MAIN, especially if these tables are small, very important tables. However, the more common case is that immediately after creating the database, the administrator creates a second main dbspace, revokes CREATE privilege IN DBSPACE IQ_SYSTEM_MAIN from all users, grants CREATE IN DBSPACE for the new main dbspace to selected users, and sets PUBLIC.default_dbspace to the new main dbspace.

• db-name.iqtmp is the default name of the file that holds the initial temporary dbspace, IQ_SYSTEM_TEMP. It contains the temporary tables generated by certain queries. The required size of this file can vary depending on the type of query and amount of data. You can specify a different name using the TEMPORARY PATH clause. This initial dbspace contains the temporary store.

• db-name.iqmsg is the default name of the file that contains the messages trace dbspace, IQ_SYSTEM_MSG. You can specify a different file name using the MESSAGE PATH clause.

In addition to these files, an IQ database has a transaction log file (db-name.log), and might have a transaction log mirror file.

The file names (db-name, log-file-name, mirror-file-name, iq-file-name, message-file-name, temp-file-name) are strings containing operating system file names. As literal strings, they must be enclosed in single quotes.

• In Windows, if you specify a path, any backslash characters (\) must be doubled if they are followed by an n or an x. This prevents them being interpreted as a newline character (\n) or as a hexadecimal number (\x), according to the rules for strings in SQL. It is safer to always double the backslash. For example:

  CREATE DATABASE 'c:\\sybase\\mydb.db'
  LOG ON 'e:\\logdrive\\mydb.log'
  JCONNECT OFF
  IQ PATH 'c:\\sybase\\mydb'
  IQ SIZE 40
  

• If you specify no path, or a relative path:

  • The catalog store file (db-name.db) is created relative to the working directory of the server.

  • The IQ store, temporary store, and message log files are created in the same directory as, or relative to, the catalog store.

Relative path names are recommended.

On UNIX systems, you can create symbolic links, which are indirect pointers that contain the path name of the file to which they point. You can use symbolic links as relative path names. There are several advantages to creating a symbolic link for the database file name:

• Symbolic links to raw devices can have meaningful names, while the actual device name syntax can be obscure.

• A symbolic name might eliminate problems restoring a database file that was moved to a new directory since it was backed up.

To create a symbolic link, use the ln -s command. For example:

ln -s /disk1/company/iqdata/company.iq company_iq_store

Once you create this link, you can specify the symbolic link in commands like CREATE DATABASE or RESTORE instead of the fully qualified path name.

When you create a database or a dbspace, the path for every dbspace file must be unique. If your CREATE DATABASE command specifies the identical path and file name for these two stores, you receive an error.

You can create a unique path in any of these ways:

• Specify a different extension for each file (for example, mydb.iq and mydb.iqtmp)

• Specify a different file name (for example, mydb.iq and mytmp.iq)

• Specify a different path name (for example, /iqfiles/main/iq and /iqfiles/temp/iq) or different raw partitions

• Omit TEMPORARY PATH when you create the database. In this case, the temporary store is created in the same path as the catalog store, with the default name and extension dbname.iqtmp, where dbname is the database name.

The default (RESPECT) is that all comparisons are case sensitive. CASE RESPECT provides better performance than CASE IGNORE.

Character strings inserted into tables are always stored in the case they are entered, regardless of whether the database is case sensitive or not. If the string Value is inserted into a character data type column, the string is always stored in the database with an uppercase V and the remainder of the letters lowercase. SELECT statements return the string as Value. If the database is not case sensitive, however, all comparisons make Value the same as value, VALUE, and so on. The IQ server may return results in any combination of lowercase and uppercase, so you cannot expect case sensitive results in a database that is case insensitive (CASE IGNORE).

For example, given the following table and data:

CREATE TABLE tb (id int NOT NULL,
                 string VARCHAR(30) NOT NULL);
INSERT INTO tb VALUES (1, ‘ONE’);
SELECT * FROM tb WHERE string = ‘oNe’;

The result of the SELECT can be ‘oNe’ (as specified in the WHERE clause) and not necessarily ‘ONE’ (as stored in the database).

Similarly, the result of

SELECT * FROM tb WHERE string = ‘One’;

can be ‘One’ and the result of

SELECT * FROM tb WHERE string = ‘ONe’;

can be ‘ONe’.

All databases are created with at least one user ID:

DBA

and password:

sql

In new databases, all passwords are case sensitive, regardless of the case-sensitivity of the database. The user ID is unaffected by the CASE RESPECT setting.

When you start a database, its page size cannot be larger than the page size of the current server. The server page size is taken from the first set of databases started or is set on the server command line using the -gp command line option.

Command line length for any statement is limited to the Catalog page size. The 4KB default is large enough in most cases; however, in a few cases a larger PAGE SIZE value is needed to accommodate very long commands, such as RESTORE commands that reference numerous dbspaces. A larger page size might also be needed to execute queries involving large numbers of tables or views.

Because the default Catalog page size is 4KB, this is only a problem when the connection is to a database such as utility_db, which has a page size of 1024. This restriction may cause RESTORE commands that reference numerous dbspaces to fail. To avoid the problem, make sure the length of SQL command lines is less than the Catalog page size.

Alternatively, start the engine with -gp 32768 to increase Catalog page size.

For most operating systems, the default collation sequence is ISO_BINENG, which provides the best performance. In ISO_BINENG, the collation order is the same as the order of characters in the ASCII character set. All uppercase letters precede all lowercase letters (for example, both ‘A’ and ‘B’ precede ‘a’).

The collation can be chosen from a list of supported collations. For SQL Anywhere databases created on a Sybase IQ server, the collation can also be the Unicode Collation Algorithm (UCA). If UCA is specified, you should also specify the ENCODING clause. For more information on the ENCODING clause, see “CREATE DATABASE statement” in SQL Anywhere Server – SQL Reference > Using SQL > SQL statements > SQL statements (A-D).

Sybase IQ does not support any of the UCA based collations for IQ databases. If a UCA based collation is specified in the CREATE DATABASE statement for an IQ database, the server returns the error “UCA collation is not supported” and database creation fails.

For syntax and a complete list of the collation tailoring options supported when specifying the UCA collation for a SQL Anywhere database, see “CREATE DATABASE statement” in SQL Anywhere Server – SQL Reference > Using SQL > SQL statements > SQL statements (A-D).

There are two levels of database and table encryption: simple and strong.

• Simple encryption is equivalent to obfuscation. The data is unreadable, but someone with cryptographic expertise could decipher the data. For simple encryption, specify the CREATE DATABASE clause ENCRYPTED ON ALGORITHM ‘SIMPLE’, ENCRYPTED ALGORITHM ‘SIMPLE’, or specify the ENCRYPTED ON clause without specifying an algorithm or key.

• Strong encryption is achieved through the use of a 128-bit algorithm and a security key. The data is unreadable and virtually undecipherable without the key. For strong encryption, specify the CREATE DATABASE clause ENCRYPTED ON ALGORITHM with a 128-bit or 256-bit AES algorithm and use the KEY clause to specify an encryption key. You should choose a value for your key that is at least 16 characters long, contains a mix of uppercase and lowercase, and includes numbers, letters, and special characters.

  This encryption key is required each time you start the database.

Encryption can be specified only during database creation. To introduce encryption to an existing database requires a complete unload, database recreation, and reload of all data.

If the ENCRYPTED clause is used but no algorithm is specified, the default is AES. Encryption is OFF by default.

For example, these two strings are treated as equal in a database created with BLANK PADDING ON:

'Smith'
'Smith   '

For the best performance, Sybase recommends the following minimum IQ page sizes:

• 64KB (IQ PAGE SIZE 65536) for databases whose largest table contains up to 1 billion rows, or a total size less than 8TB. This is the absolute minimum for a new database. On 32-bit platforms, a 64KB IQ page size gives the best performance.

• 128KB (IQ PAGE SIZE 131072) for databases on a 64-bit platform whose largest table contains more than 1 billion rows and fewer than 4 billion rows, or might grow to a total size of 8TB or greater. 128KB is the default IQ page size.

• 256KB (IQ PAGE SIZE 262144) for databases on a 64-bit platform whose largest table contains more than 4 billion rows, or might grow to a total size of 8TB or greater.

Very wide tables, such as tables with multiple columns of wide VARCHAR data (columns from 255 to 32,767 bytes) might need the next larger IQ PAGE SIZE.

When IQ RESERVE is specified, the database uses more space for internal (free list) structures. If reserve size is too large, the space needed for the internal structures can be larger than the specified size, which results in an error.

When TEMPORARY RESERVE is specified, the database uses more space for internal (free list) structures. If reserve size is too large, the space needed for the internal structures can be larger than the specified size, which results in an error.

Several operating system files are created.

Standards

Permissions

The permissions required to execute this statement are set on the server command line, using the -gu option. The default setting is to require DBA authority.

The account under which the server is running must have write permissions on the directories where files are created.

See also