Teradata Informatica Best Practices
Content
Teradata Best Practices with Informatica PowerCenter
Introduction This document discusses configuration and supplies how-to examples for using Informatica PowerCenter® 7.1.2 and NCR Teradata RDBMS. It covers Teradata basics and also describes some “tweaks” that experience has shown may be necessary to adequately deal with some of the common practices you may encounter at a Teradata account. The Teradata documentation (especially the MultiLoad, FastLoad, and TPump reference) is highly recommended reading material, as is the “External Loader” section of the Server Manager Guide for PowerCenter. Additional Information: All Teradata documentation can be downloaded from the NCR Web site (http://www.info.ncr.com/Teradata/eTeradata-BrowseBy.cfm). Finally, a “Teradata Forum” provides a wealth of information that can be useful (http://www.Teradataforum.com). Teradata Basics Teradata is a relational database management system from NCR. It offers high performance for very large database tables because of its highly parallel architecture. It is a major player in the retail space. Although Teradata can run on other platforms, it predominantly appears on NCR hardware (which runs NCR’s version of UNIX). Teradata is very fast and very scalable. Teradata Hardware The NCR computers on which Teradata runs support both massively parallel processing (MPP) and symmetric multiprocessing (SMP). Each MPP “node” (or semi-autonomous processing unit) can support SMP. Teradata can be configured to communicate directly with a mainframe’s input/output (I/O) channel, known as “channel attached.” Alternatively, it can be “network attached,” configured to communicate via transmission control protocol/Internet protocol (TCP/IP) over a local area network (LAN). Because PowerCenter runs on UNIX, you will be dealing with a network-attached configuration most of the time. However, once in a while, clients will want to use their existing channel-attached configuration under the auspices of better performance. Do not assume that channel attached is always faster than network attached. Similar performance has been observed across a channel attachment as well as a 100-MB LAN. In addition, channel attachment requires an additional sequential data move: Data must be moved from the PowerCenter server to the mainframe before moving the data across the mainframe channel to Teradata. Teradata Software In the Teradata world, there are Teradata Director Program IDs (TDPIDs), databases, and users. The TDPID is simply the name for connections from a Teradata client to Teradata server (similar to an Oracle “tnsnames.ora” entry). Teradata also looks at databases and users somewhat synonymously. A user has a userid, password, and space to store tables. A database is basically a user without a login and password (or a user is a database with a userid and password). Teradata AMPs are access module processors. Think of them as Teradata’s parallel database engines. Although they are strictly software (“virtual processors” according to NCR terminology), Teradata employees often use AMP and hardware “node” interchangeably because an AMP previously was a piece of hardware. Client Configuration Basics for Teradata The client-side configuration is wholly contained in the “hosts” file (/etc/hosts on UNIX or winnt\system32\drivers\etc\hosts on Win). Informatica does not run on NCR UNIX, so you should not have to deal with the server side. Teradata uses a naming nomenclature in the “hosts” file. The name of the Teradata instance (that is, TDPID) is indicated by the letters and numbers that precede the string “cop1” in a “hosts” file entry. For example: 127.0.0.1 192.168.80.113 localhost curly demo1099cop1 pcop1
This tells Teradata that when a client tool references the instance “demo1099,” it should direct requests to “localhost” (or IP address 127.0.0.1); when a client tool references instance “p”, it is located on the server “curly” (or IP address 192.168.80.113). There is no tie here to any kind of database server specific information (This is not similar to Oracle’s instance id. TDPID <> Oracle instance id.). That is, the TDPID is used strictly to define the name a client uses to connect to a server. You can call a server whatever you want. Teradata does not
2 Informatica Confidential. Do not duplicate.
care. It simply takes the name you specify, looks in the “hosts” file to map the <name>cop1 (or cop2, and so on) to an IP address, and then attempts to establish a connect with Teradata at the IP address. Sometimes you’ll see multiple entries in a “hosts” file with similar TDPIDs: 127.0.0.1 192.168.80.113 192.168.80.114 192.168.80.115 192.168.80.116 localhost curly_1 curly_2 curly_3 curly_4 demo1099cop1 pcop1 pcop2 pcop3 pcop4
This setup allows load balancing of clients among multiple Teradata nodes. That is, most Teradata systems have many nodes, and each node has its own IP address. Without the multiple “hosts” file entries, every client will connect to one node and eventually this node will be doing more than its fair share of client processing. With multiple “hosts” file entries, if it takes too long for the node specified with the “cop1” suffix to respond (that is, curly_1) to the client request to connect to “p”, then the client will automatically attempt to connect to the node with the “cop2” suffix (that is, curly_2). Informatica/Teradata Touch Points Informatica PowerCenter 7.1.2 accesses Teradata through various Teradata tools. Each will be defined according to how it is configured within PowerCenter. ODBC: Teradata provides 32-bit ODBC drivers for Windows and UNIX platforms. If possible, use the ODBC driver from Teradata’s TTU7 release (or above) of its client software because this version supports “array reads”. Tests have shown these “new” drivers (3.02) can be 20 to 30 percent faster than the “old” drivers (3.01). This latest release of Teradata’s TTU8.0 uses ODBC v3.0421. Teradata’s ODBC is on a performance par with Teradata’s SQL CLI. In fact, ODBC is Teradata’s recommended SQL interface for its partners. Do not use ODBC to write to Teradata unless you’re writing very small data sets (and even then, you should probably use TPump, defined later, instead) because Teradata’s ODBC is optimized for query access, not for writing data. ODBC is good for sourcing and lookups. PowerCenter Designer uses Teradata’s ODBC to import Source and Target table. If you are having performance problems, you can use a cmd task with a shell script to call BTEQ. A SQL with an intermediate work table can be sourced by PowerCenter.
3 Informatica Confidential. Do not duplicate.
ODBC Windows:
ODBC UNIX When the PowerCenter server is running on UNIX, then ODBC is required to read (both sourcing and lookups) from Teradata. As with all UNIX ODBC drivers, the key to configuring the UNIX ODBC driver is adding the appropriate entries to the “.odbc.ini” file. To correctly configure the “.odbc.ini” file, there must be an entry under [ODBC Data Sources] that points to the Teradata ODBC driver shared library (tdata.sl on HP-UX, the standard shared library extension on other flavors of UNIX). The following example shows the required entries from an actual “.odbc.ini” file (note that the path to the driver may be different on each computer): [ODBC Data Sources] dBase=MERANT 3.60 dBase Driver Oracle8=MERANT 3.60 Oracle 8 Driver Text=MERANT 3.60 Text Driver Sybase11=MERANT 3.60 Sybase 11 Driver Informix=MERANT 3.60 Informix Driver DB2=MERANT 3.60 DB2 Driver MS_SQLServer7=MERANT SQLServer driver
4 Informatica Confidential. Do not duplicate.
TeraTest=tdata.sl [TeraTest] Driver=/usr/odbc/drivers/tdata.sl Description=Teradata Test System DBCName=148.162.247.34 Similar to the client “hosts” file setup, you can specify multiple IP addresses for the DBCName to balance the client load across multiple Teradata nodes. Consult with the Teradata administrator for exact details on this (or copy the entries from the PC client’s “hosts” file (see the section “Client Configuration Basics for Teradata” earlier in this document). Important note: Make sure that the Merant ODBC path precedes the Teradata ODBC path information in the PATH and SHLIB_PATH (or LD_LIBRARY_PATH, and so on) environment variables. This is necessary because both sets of ODBC software use some of the same filenames. PowerCenter should use the Merant files because this software has been certified. Teradata External Loaders PowerCenter 7.1.2 supports four different Teradata external loaders: TPump, FastLoad, MultiLoad, and Teradata Warehouse Builder (TWB). The actual Teradata loader executables (TPump, mload, fastload, tbuild) must be accessible by the PowerCenter server generally in the path statement. All of the Teradata loader connections will require a value to the TDPID attribute. Refer to the first section of this document to understand how to correctly enter the value. All of these loaders require: • • A load file, which can be configured to be a stream/pipe and is autogenerated by PowerCenter A control file of commands to tell the loader what to do (PowerCenter autogenerates)
All of these loaders will also produce a log file, which will be the means to debug the loader if something goes wrong. Because these are external loaders, PowerCenter will only receive back from the loader whether it ran successfully or not.
5 Informatica Confidential. Do not duplicate.
By default, the input file, control file, and log file will be created in $PMTargetFileDir of the PowerCenter server executing the workflow.
6 Informatica Confidential. Do not duplicate.
You can use any of these loaders by configuring the target in the PowerCenter session to be a “File Writer” and then choosing the appropriate loader.
7 Informatica Confidential. Do not duplicate.
The autogenerated control file can be overridden. Click the Pencil icon next to the loader connection name.
8 Informatica Confidential. Do not duplicate.
Scroll to the bottom of the connection attribute list and click the value next to the “Control File Content Override” attribute. Then click the Down arrow.
9 Informatica Confidential. Do not duplicate.
Click the Generate button and change the control file as you wish. The repository stores the changed control file.
10 Informatica Confidential. Do not duplicate.
Most of the loaders also use some combination of internal work, error, and log tables. By default, these will be in the same database as the target table. All of these can now be overridden in the attributes of the connection.
To land the input flat file to disk, check the “Is Staged” attribute. If the “Is Staged” attribute is not set, then the file will be piped/streamed to the loader. If you select the nonstaged mode for a loader, also set the “checkpoint” property to 0. This effectively turns off the “checkpoint” processing. “Checkpoint” processing is used for recovery/restart of FastLoad and MultiLoad sessions. However, if you are using a named pipe instead of a physical file as input, then the recovery/restart mechanism of the loaders does not work. Besides impacting performance (the checkpoint processing is not free and we want to eliminate unnecessary overhead when possible), a nonzero checkpoint value will sometimes cause seemingly random errors and session failures when used with named pipe input (as is the case with “streaming” mode). Teradata Loader Requirements for PowerCenter Servers on UNIX All Teradata load utilities require a “non-null” standard output and standard error to run properly. “Standard output” (stdout) and “standard error” (stderr) are UNIX conventions that determine the default location for a program to write output and error information. When you start the pmserver without explicitly defining stdout and stderr, these both point to the current terminal session. If you log out of UNIX, then UNIX redirects stdout and stderr to /dev/null (that is, a placeholder that throws out anything written to it). At this point, Teradata loader
11 Informatica Confidential. Do not duplicate.
sessions will fail because they do not permit stdout and stderr to be /dev/null. Therefore, you must start pmserver as follows (“cd” to the PowerCenter installation directory): ./pmserver ./pmserver.cfg > ./pmserver.out 2>&1 This starts the pmserver using the “pmserver.cfg” config file and points stdout and stderr to the file “pmserver.out”. In this way, stderr and stdout will be defined even after the terminal session logs out. Important note: There are no spaces in the token “2>&1”. This tells UNIX to point stderr to the same place stdout is pointing. As an alternative to this method, you can specify the console output filename in the pmserver.cfg file. That is, information written to “standard output” and “standard error” will go the file specified as follows: ConsoleOutputFilename=<FILE_NAME> With this entry in the pmserver.cfg file, you can start the pmserver normally (that is, ./pmserver). Partitioned Loading With PowerCenter v7.x, if you set a “round robin” partition point on the target definition and set each target instance to be loaded using the same loader connection instance, then PowerCenter automatically writes all data to the first partition and only starts one instance of FastLoad or MultiLoad. You will know you are getting this behavior if you see the following entry in the session log: MAPPING> DBG_21684 Target [TD_INVENTORY] does not support multiple partitions. All data will be routed to the first partition. If you do not see this message, then chances are the session fails with the following error: WRITER_1_*_1> WRT_8240 Error: The external loader [Teradata Mload Loader] does not support partitioned sessions. WRITER_1_*_1> Thu Jun 16 11:58:21 2005 WRITER_1_*_1> WRT_8068 Writer initialization failed. Writer terminating.
12 Informatica Confidential. Do not duplicate.
TPump TPump is an external loader that supports inserts, updates, upserts, deletes, and data-driven updates. Multiple TPumps can execute simultaneously against the same table because TPump doesn’t use many resources or require table-level locks. It is often used to “trickle load” a table. As stated earlier, TPump will be a faster way to update a table than using ODBC, but will not be as fast as the other loaders.
13 Informatica Confidential. Do not duplicate.
MultiLoad This sophisticated bulk load utility is the primary method PowerCenter uses to load/update mass quantities of data into Teradata. Unlike bulk load utilities from other vendors, MultiLoad supports inserts, updates, upserts, deletes, and data-driven operations in PowerCenter. You can also use variables and embed conditional logic into MultiLoad scripts. It is very fast (millions of rows in a few minutes). It can be resource intensive and will take a table lock.
Cleaning up after a failed MultiLoad: MultiLoad supports sophisticated error recovery. That is, it allows load jobs to be restarted without having to redo all of the prior work. However, for the types of problems normally encountered during a POC (loading null values into a column that does not support nulls, incorrectly formatted date columns), the error recovery mechanisms tend to get in the way. To learn about MultiLoad’s sophisticated error recovery, read the Teradata MultiLoad manual. To learn how to work around the recovery mechanisms to restart a failed MultiLoad script from scratch, read this section.
14 Informatica Confidential. Do not duplicate.
MultiLoad puts the target table into the “MultiLoad” state. Upon successful completion, the target table is returned to the “normal” (non“MultiLoad”) state. Therefore, when a MultiLoad fails for any reason, the table is left in the “MultiLoad” state, and you cannot simply rerun the same MultiLoad. MultiLoad will report an error. In addition, MultiLoad also queries the target table’s MultiLoad log table to see if it contains any errors. If a MultiLoad log table exists for the target table, then you also will not be able to rerun your MultiLoad job. To recover from a failed MultiLoad, “release” the target table from the “MultiLoad” state and also drop the MultiLoad log table. You can do this using BTEQ or QueryMan to issue the following commands: drop table mldlog_<table name>; release mload <table name>; Note: The “drop table” command assumes that you’re recovering from a MultiLoad script generated by PowerCenter (PowerCenter always names the MultiLoad log table “mldlog_<table name>”). If you’re working with a hand-coded MultiLoad script, the name of the MultiLoad log table could be anything. Here is the actual text from a BTEQ session that cleans up a failed load to the table “td_test” owned by the user “infatest:” BTEQ -- Enter your DBC/SQL request or BTEQ command: drop table infatest.mldlog_td_test; drop table infatest.mldlog_td_test; *** Table has been dropped. *** Total elapsed time was 1 second. BTEQ -- Enter your DBC/SQL request or BTEQ command: release mload infatest.td_test; release mload infatest.td_test; *** Mload has been released. *** Total elapsed time was 1 second. Using One Instance of MultiLoad to Load Multiple Tables MultiLoad is a big consumer of resources on a Teradata system. Some systems will have hard limits on the number of concurrent MultiLoad sessions allowed. By default, PowerCenter will start an instance of MultiLoad for every target file. Sometimes, this is illegal (if the multiple instances target the same table). Other times, it is just expensive. Therefore, a prospect may ask that PowerCenter use a single instance of MultiLoad to load multiple tables (or to load both inserts and updates into the same target table). To make this happen, we must heavily edit the generated MultiLoad script file. Note: This should not be an issue with TPump because TPump is not as resource intensive as MultiLoad (and a multiple concurrent instances of TPump can target the same table). Here’s the workaround: 1) 2) 3) Use a dummy session (that is, set test rows to 1 and target a test database) to generate MultiLoad control files for each of the targets. Merge the multiple control files (one per target table) into a single control file (one for all target tables). Configure the session to call MultiLoad from a postsession script using the control file created in step 2. Integrated support cannot be used because each input file is processed sequentially and this causes problems when combined with the integrated named pipes and streaming of PowerCenter.
Details on “merging” the control files: 1) There is a single log file for each instance of MultiLoad. Therefore, you do not have to change or add anything the “LOGFILE” statement. However, you might want to change the name of the log table because it may be a log that spans multiple tables.
15 Informatica Confidential. Do not duplicate.
2) 3) 4) 5) 6) 7) 8)
Copy the work and error tables’ delete statements into the common control file. Modify the “BEGIN MLOAD” statement to specify all the tables that the MultiLoad will be hitting. Copy the “Layout” sections into the common control file and give each a unique name. Organize the file such that all the layout sections are grouped together. Copy the “DML” sections into the common control file and give each a unique name. Organize the file such that all the DML sections are grouped together. Copy the “Import” statements into the common control file and modify them to reflect the unique names created for the referenced layout and DML sections created in steps 4 and 5. Organize the file such that all the import sections are grouped together. Run “chmod –w” on the newly minted control file so PowerCenter doesn’t overwrite it, or, better yet, name it something different so PowerCenter cannot overwrite it. Remember, a single instance of MultiLoad can target at most five tables. Therefore, don’t combine more than five target files into a common file.
Here’s an example of a control file merged from two default control files: .DATEFORM ANSIDATE; .LOGON demo1099/infatest,infatest; .LOGTABLE infatest.mldlog_TD_TEST; DROP TABLE infatest.UV_TD_TEST ; DROP TABLE infatest.WT_TD_TEST ; DROP TABLE infatest.ET_TD_TEST ; DROP TABLE infatest.UV_TD_CUSTOMERS ; DROP TABLE infatest.WT_TD_CUSTOMERS ; DROP TABLE infatest.ET_TD_CUSTOMERS ; .ROUTE MESSAGES WITH ECHO TO FILE c:\LOGS\TgtFiles\td_test.out.ldrlog ; .BEGIN IMPORT MLOAD TABLES infatest.TD_TEST, infatest.TD_CUSTOMERS ERRLIMIT 1 CHECKPOINT 10000 TENACITY 10000 SESSIONS 1 SLEEP 6 ; /* Begin Layout Section */ .Layout InputFileLayout1; .Field .Field .Field .Field .Field .Field CUST_KEY 1 CHAR( 12) NULLIF CUST_KEY = '*' ; CUST_NAME 13 CHAR( 20) NULLIF CUST_NAME = '*' ; CUST_DATE 33 CHAR( 10) NULLIF CUST_DATE = '*' ; CUST_DATEmm 33 CHAR( 2) ; CUST_DATEdd 36 CHAR( 2) ; CUST_DATEyyyy 39 CHAR( 4) ;
16 Informatica Confidential. Do not duplicate.
.Field CUST_DATEtd CUST_DATE .Filler EOL_PAD
CUST_DATEyyyy||'/'||CUST_DATEmm||'/'||CUST_DATEdd NULLIF = '*' ; 43 CHAR( 2) ;
.Layout InputFileLayout2; .Field CUSTOMER_KEY 1 CHAR( 12) ; .Field CUSTOMER_ID 13 CHAR( 12) ; .Field COMPANY 25 CHAR( 50) NULLIF COMPANY = '*' ; .Field FIRST_NAME 75 CHAR( 30) NULLIF FIRST_NAME = '*' ; .Field LAST_NAME 105 CHAR( 30) NULLIF LAST_NAME = '*' ; .Field ADDRESS1 135 CHAR( 72) NULLIF ADDRESS1 = '*' ; .Field ADDRESS2 207 CHAR( 72) NULLIF ADDRESS2 = '*' ; .Field CITY 279 CHAR( 30) NULLIF CITY = '*' ; .Field STATE 309 CHAR( 2) NULLIF STATE = '*' ; .Field POSTAL_CODE 311 CHAR( 10) NULLIF POSTAL_CODE = '*' ; .Field PHONE 321 CHAR( 30) NULLIF PHONE = '*' ; .Field EMAIL 351 CHAR( 30) NULLIF EMAIL = '*' ; .Field REC_STATUS 381 CHAR( 1) NULLIF REC_STATUS = '*' ; .Filler EOL_PAD 382 CHAR( 2) ; /* End Layout Section */ /* begin DML Section */ .DML Label tagDML1; INSERT INTO infatest.TD_TEST ( CUST_KEY , CUST_NAME , CUST_DATE ) VALUES ( :CUST_KEY , :CUST_NAME , :CUST_DATEtd ) ; .DML Label tagDML2; INSERT INTO infatest.TD_CUSTOMERS ( CUSTOMER_KEY , CUSTOMER_ID , COMPANY , FIRST_NAME , LAST_NAME , ADDRESS1 , ADDRESS2 , CITY , STATE , POSTAL_CODE , PHONE , EMAIL , REC_STATUS ) VALUES (
17 Informatica Confidential. Do not duplicate.
:CUSTOMER_KEY :CUSTOMER_ID :COMPANY :FIRST_NAME :LAST_NAME :ADDRESS1 :ADDRESS2 :CITY :STATE :POSTAL_CODE :PHONE :EMAIL :REC_STATUS ) ; /* end DML Section */
,
, ,
, , , ,
, , , , ,
/* Begin Import Section */ .Import Infile c:\LOGS\TgtFiles\td_test.out Layout InputFileLayout1 Format Unformat Apply tagDML1 ; .Import Infile c:\LOGS\TgtFiles\td_customers.out Layout InputFileLayout2 Format Unformat Apply tagDML2 ; /* End Import Section */ .END MLOAD; .LOGOFF;
18 Informatica Confidential. Do not duplicate.
Multiple Workflows That MultiLoad to the Same Table Because MultiLoad puts a lock on the table, we require that all MultiLoad sessions must handle wait events so they don't try to access the table simultaneously. Also, any log files should be given unique names for the same reason. FastLoad As the name suggests, this utility is the fastest method to load data into Teradata. However, there is one major restriction: the target table must be empty.
19 Informatica Confidential. Do not duplicate.
Teradata Warehouse Builder (TWB) Teradata Warehouse Builder (TWB) is a single utility that was intended to replace FastLoad, MultiLoad, TPump, and FastExport. It was to support a single scripting environment with different “modes,” where each “mode” roughly equates to one of the legacy utilities. It also was to support parallel loading (that is, multiple instances of a TWB client could run and load the same table at the same time—something the legacy loaders cannot do). Although PowerCenter supports TWB, NCR/Teradata does not. TWB has never been formally released. According to NCR, its general release was delayed primarily because of issues with the mainframe version. If you find a prospect willing to use TWB, please do. Its ability to support parallel load clients makes some tasks easier.
20 Informatica Confidential. Do not duplicate.
PowerCenter 7.1.2 PAM for Teradata Server Client Software Platform Version & Version Source Target
Comment
PowerCenter uses the following components of the Teradata Tools and Utilities (TTU): Teradata ODBC driver, FastLoad, MultiLoad, and TPump. The version numbers of each of the Teradata Client components vary with release (e.g., TTU7 contains ODBC 3.02, FastLoad 7.05, MultiLoad 3.03, TPump 1.07). The TTU was previously called Teradata Utilities Foundation (TUF). Compatibility between a particular version of the Teradata RDBMS and the Teradata Client software is determined by Teradata -- not Informatica.
NCR Teradata
UNIX and NT
V2R4
TUF 6.1, TTU 6.1.1, TTU 6.2
x
x
UNIX and NT
TUF 6.1, TTU 6.1.1, V2R4.1 TTU 6.2, TTU 7
x
x The Teradata Client/Teradata RDBMS pairings listed here represent our understanding based on Teradata's documentation. Note that the minimum version number for the Teradata ODBC driver is 3.00.01.04. Teradata has made many fixes to the 3.02 ODBC driver. If you are using this driver, please contact NCR support for the latest maintenance release.
UNIX and NT
V2R5
TTU 7
x
x
UNIX and NT
V2R5.1 TTU 7
x
x
UNIX and NT
V2R6
TTU 8
x
x
Supported Teradata ODBC driver is v3.0.4.1.
21 Informatica Confidential. Do not duplicate.
New PowerCenter Features Not Covered in this Document
FastExport As its name implies, FastExport is a utility to extract data from Teradata very quickly. It is supported in PowerCenter 7.1.3 for Teradata sources. This feature will extract large tables used for lookups quicker than ODBC,. Pushdown Optimization The upcoming release of PowerCenter 8 will be feature a Pushdown Optimization Option. This option enables data transformation processing, where appropriate, to be “pushed down” into a relational database. Powered by an easy-to-use, visual user interface, the option streamlines the generation and execution of platform-specific business logic and manages the execution in a database as opposed to the Informatica Data Server. The option is particularly useful in scenarios where data is “co-resident” (i.e., staging and production are in the same database instance). With this new functionality, users will be able to create SQL that will execute in the Teradata database server, replacing certain transformations that would normally run in the Informatica Data Server. This is critical to Teradata because the Source and Target commonly reside in the same database.
©Copyright 2005 Informatica Corporation. Informatica and PowerCenter are registered trademarks of Informatica Corporation. Teradata is a registered trademark of NCR Corporation. All other company, product, or service names may be trademarks or registered trademarks of their respective owners.
J50665 (9/29/2005)
22 Informatica Confidential. Do not duplicate.
Introduction This document discusses configuration and supplies how-to examples for using Informatica PowerCenter® 7.1.2 and NCR Teradata RDBMS. It covers Teradata basics and also describes some “tweaks” that experience has shown may be necessary to adequately deal with some of the common practices you may encounter at a Teradata account. The Teradata documentation (especially the MultiLoad, FastLoad, and TPump reference) is highly recommended reading material, as is the “External Loader” section of the Server Manager Guide for PowerCenter. Additional Information: All Teradata documentation can be downloaded from the NCR Web site (http://www.info.ncr.com/Teradata/eTeradata-BrowseBy.cfm). Finally, a “Teradata Forum” provides a wealth of information that can be useful (http://www.Teradataforum.com). Teradata Basics Teradata is a relational database management system from NCR. It offers high performance for very large database tables because of its highly parallel architecture. It is a major player in the retail space. Although Teradata can run on other platforms, it predominantly appears on NCR hardware (which runs NCR’s version of UNIX). Teradata is very fast and very scalable. Teradata Hardware The NCR computers on which Teradata runs support both massively parallel processing (MPP) and symmetric multiprocessing (SMP). Each MPP “node” (or semi-autonomous processing unit) can support SMP. Teradata can be configured to communicate directly with a mainframe’s input/output (I/O) channel, known as “channel attached.” Alternatively, it can be “network attached,” configured to communicate via transmission control protocol/Internet protocol (TCP/IP) over a local area network (LAN). Because PowerCenter runs on UNIX, you will be dealing with a network-attached configuration most of the time. However, once in a while, clients will want to use their existing channel-attached configuration under the auspices of better performance. Do not assume that channel attached is always faster than network attached. Similar performance has been observed across a channel attachment as well as a 100-MB LAN. In addition, channel attachment requires an additional sequential data move: Data must be moved from the PowerCenter server to the mainframe before moving the data across the mainframe channel to Teradata. Teradata Software In the Teradata world, there are Teradata Director Program IDs (TDPIDs), databases, and users. The TDPID is simply the name for connections from a Teradata client to Teradata server (similar to an Oracle “tnsnames.ora” entry). Teradata also looks at databases and users somewhat synonymously. A user has a userid, password, and space to store tables. A database is basically a user without a login and password (or a user is a database with a userid and password). Teradata AMPs are access module processors. Think of them as Teradata’s parallel database engines. Although they are strictly software (“virtual processors” according to NCR terminology), Teradata employees often use AMP and hardware “node” interchangeably because an AMP previously was a piece of hardware. Client Configuration Basics for Teradata The client-side configuration is wholly contained in the “hosts” file (/etc/hosts on UNIX or winnt\system32\drivers\etc\hosts on Win). Informatica does not run on NCR UNIX, so you should not have to deal with the server side. Teradata uses a naming nomenclature in the “hosts” file. The name of the Teradata instance (that is, TDPID) is indicated by the letters and numbers that precede the string “cop1” in a “hosts” file entry. For example: 127.0.0.1 192.168.80.113 localhost curly demo1099cop1 pcop1
This tells Teradata that when a client tool references the instance “demo1099,” it should direct requests to “localhost” (or IP address 127.0.0.1); when a client tool references instance “p”, it is located on the server “curly” (or IP address 192.168.80.113). There is no tie here to any kind of database server specific information (This is not similar to Oracle’s instance id. TDPID <> Oracle instance id.). That is, the TDPID is used strictly to define the name a client uses to connect to a server. You can call a server whatever you want. Teradata does not
2 Informatica Confidential. Do not duplicate.
care. It simply takes the name you specify, looks in the “hosts” file to map the <name>cop1 (or cop2, and so on) to an IP address, and then attempts to establish a connect with Teradata at the IP address. Sometimes you’ll see multiple entries in a “hosts” file with similar TDPIDs: 127.0.0.1 192.168.80.113 192.168.80.114 192.168.80.115 192.168.80.116 localhost curly_1 curly_2 curly_3 curly_4 demo1099cop1 pcop1 pcop2 pcop3 pcop4
This setup allows load balancing of clients among multiple Teradata nodes. That is, most Teradata systems have many nodes, and each node has its own IP address. Without the multiple “hosts” file entries, every client will connect to one node and eventually this node will be doing more than its fair share of client processing. With multiple “hosts” file entries, if it takes too long for the node specified with the “cop1” suffix to respond (that is, curly_1) to the client request to connect to “p”, then the client will automatically attempt to connect to the node with the “cop2” suffix (that is, curly_2). Informatica/Teradata Touch Points Informatica PowerCenter 7.1.2 accesses Teradata through various Teradata tools. Each will be defined according to how it is configured within PowerCenter. ODBC: Teradata provides 32-bit ODBC drivers for Windows and UNIX platforms. If possible, use the ODBC driver from Teradata’s TTU7 release (or above) of its client software because this version supports “array reads”. Tests have shown these “new” drivers (3.02) can be 20 to 30 percent faster than the “old” drivers (3.01). This latest release of Teradata’s TTU8.0 uses ODBC v3.0421. Teradata’s ODBC is on a performance par with Teradata’s SQL CLI. In fact, ODBC is Teradata’s recommended SQL interface for its partners. Do not use ODBC to write to Teradata unless you’re writing very small data sets (and even then, you should probably use TPump, defined later, instead) because Teradata’s ODBC is optimized for query access, not for writing data. ODBC is good for sourcing and lookups. PowerCenter Designer uses Teradata’s ODBC to import Source and Target table. If you are having performance problems, you can use a cmd task with a shell script to call BTEQ. A SQL with an intermediate work table can be sourced by PowerCenter.
3 Informatica Confidential. Do not duplicate.
ODBC Windows:
ODBC UNIX When the PowerCenter server is running on UNIX, then ODBC is required to read (both sourcing and lookups) from Teradata. As with all UNIX ODBC drivers, the key to configuring the UNIX ODBC driver is adding the appropriate entries to the “.odbc.ini” file. To correctly configure the “.odbc.ini” file, there must be an entry under [ODBC Data Sources] that points to the Teradata ODBC driver shared library (tdata.sl on HP-UX, the standard shared library extension on other flavors of UNIX). The following example shows the required entries from an actual “.odbc.ini” file (note that the path to the driver may be different on each computer): [ODBC Data Sources] dBase=MERANT 3.60 dBase Driver Oracle8=MERANT 3.60 Oracle 8 Driver Text=MERANT 3.60 Text Driver Sybase11=MERANT 3.60 Sybase 11 Driver Informix=MERANT 3.60 Informix Driver DB2=MERANT 3.60 DB2 Driver MS_SQLServer7=MERANT SQLServer driver
4 Informatica Confidential. Do not duplicate.
TeraTest=tdata.sl [TeraTest] Driver=/usr/odbc/drivers/tdata.sl Description=Teradata Test System DBCName=148.162.247.34 Similar to the client “hosts” file setup, you can specify multiple IP addresses for the DBCName to balance the client load across multiple Teradata nodes. Consult with the Teradata administrator for exact details on this (or copy the entries from the PC client’s “hosts” file (see the section “Client Configuration Basics for Teradata” earlier in this document). Important note: Make sure that the Merant ODBC path precedes the Teradata ODBC path information in the PATH and SHLIB_PATH (or LD_LIBRARY_PATH, and so on) environment variables. This is necessary because both sets of ODBC software use some of the same filenames. PowerCenter should use the Merant files because this software has been certified. Teradata External Loaders PowerCenter 7.1.2 supports four different Teradata external loaders: TPump, FastLoad, MultiLoad, and Teradata Warehouse Builder (TWB). The actual Teradata loader executables (TPump, mload, fastload, tbuild) must be accessible by the PowerCenter server generally in the path statement. All of the Teradata loader connections will require a value to the TDPID attribute. Refer to the first section of this document to understand how to correctly enter the value. All of these loaders require: • • A load file, which can be configured to be a stream/pipe and is autogenerated by PowerCenter A control file of commands to tell the loader what to do (PowerCenter autogenerates)
All of these loaders will also produce a log file, which will be the means to debug the loader if something goes wrong. Because these are external loaders, PowerCenter will only receive back from the loader whether it ran successfully or not.
5 Informatica Confidential. Do not duplicate.
By default, the input file, control file, and log file will be created in $PMTargetFileDir of the PowerCenter server executing the workflow.
6 Informatica Confidential. Do not duplicate.
You can use any of these loaders by configuring the target in the PowerCenter session to be a “File Writer” and then choosing the appropriate loader.
7 Informatica Confidential. Do not duplicate.
The autogenerated control file can be overridden. Click the Pencil icon next to the loader connection name.
8 Informatica Confidential. Do not duplicate.
Scroll to the bottom of the connection attribute list and click the value next to the “Control File Content Override” attribute. Then click the Down arrow.
9 Informatica Confidential. Do not duplicate.
Click the Generate button and change the control file as you wish. The repository stores the changed control file.
10 Informatica Confidential. Do not duplicate.
Most of the loaders also use some combination of internal work, error, and log tables. By default, these will be in the same database as the target table. All of these can now be overridden in the attributes of the connection.
To land the input flat file to disk, check the “Is Staged” attribute. If the “Is Staged” attribute is not set, then the file will be piped/streamed to the loader. If you select the nonstaged mode for a loader, also set the “checkpoint” property to 0. This effectively turns off the “checkpoint” processing. “Checkpoint” processing is used for recovery/restart of FastLoad and MultiLoad sessions. However, if you are using a named pipe instead of a physical file as input, then the recovery/restart mechanism of the loaders does not work. Besides impacting performance (the checkpoint processing is not free and we want to eliminate unnecessary overhead when possible), a nonzero checkpoint value will sometimes cause seemingly random errors and session failures when used with named pipe input (as is the case with “streaming” mode). Teradata Loader Requirements for PowerCenter Servers on UNIX All Teradata load utilities require a “non-null” standard output and standard error to run properly. “Standard output” (stdout) and “standard error” (stderr) are UNIX conventions that determine the default location for a program to write output and error information. When you start the pmserver without explicitly defining stdout and stderr, these both point to the current terminal session. If you log out of UNIX, then UNIX redirects stdout and stderr to /dev/null (that is, a placeholder that throws out anything written to it). At this point, Teradata loader
11 Informatica Confidential. Do not duplicate.
sessions will fail because they do not permit stdout and stderr to be /dev/null. Therefore, you must start pmserver as follows (“cd” to the PowerCenter installation directory): ./pmserver ./pmserver.cfg > ./pmserver.out 2>&1 This starts the pmserver using the “pmserver.cfg” config file and points stdout and stderr to the file “pmserver.out”. In this way, stderr and stdout will be defined even after the terminal session logs out. Important note: There are no spaces in the token “2>&1”. This tells UNIX to point stderr to the same place stdout is pointing. As an alternative to this method, you can specify the console output filename in the pmserver.cfg file. That is, information written to “standard output” and “standard error” will go the file specified as follows: ConsoleOutputFilename=<FILE_NAME> With this entry in the pmserver.cfg file, you can start the pmserver normally (that is, ./pmserver). Partitioned Loading With PowerCenter v7.x, if you set a “round robin” partition point on the target definition and set each target instance to be loaded using the same loader connection instance, then PowerCenter automatically writes all data to the first partition and only starts one instance of FastLoad or MultiLoad. You will know you are getting this behavior if you see the following entry in the session log: MAPPING> DBG_21684 Target [TD_INVENTORY] does not support multiple partitions. All data will be routed to the first partition. If you do not see this message, then chances are the session fails with the following error: WRITER_1_*_1> WRT_8240 Error: The external loader [Teradata Mload Loader] does not support partitioned sessions. WRITER_1_*_1> Thu Jun 16 11:58:21 2005 WRITER_1_*_1> WRT_8068 Writer initialization failed. Writer terminating.
12 Informatica Confidential. Do not duplicate.
TPump TPump is an external loader that supports inserts, updates, upserts, deletes, and data-driven updates. Multiple TPumps can execute simultaneously against the same table because TPump doesn’t use many resources or require table-level locks. It is often used to “trickle load” a table. As stated earlier, TPump will be a faster way to update a table than using ODBC, but will not be as fast as the other loaders.
13 Informatica Confidential. Do not duplicate.
MultiLoad This sophisticated bulk load utility is the primary method PowerCenter uses to load/update mass quantities of data into Teradata. Unlike bulk load utilities from other vendors, MultiLoad supports inserts, updates, upserts, deletes, and data-driven operations in PowerCenter. You can also use variables and embed conditional logic into MultiLoad scripts. It is very fast (millions of rows in a few minutes). It can be resource intensive and will take a table lock.
Cleaning up after a failed MultiLoad: MultiLoad supports sophisticated error recovery. That is, it allows load jobs to be restarted without having to redo all of the prior work. However, for the types of problems normally encountered during a POC (loading null values into a column that does not support nulls, incorrectly formatted date columns), the error recovery mechanisms tend to get in the way. To learn about MultiLoad’s sophisticated error recovery, read the Teradata MultiLoad manual. To learn how to work around the recovery mechanisms to restart a failed MultiLoad script from scratch, read this section.
14 Informatica Confidential. Do not duplicate.
MultiLoad puts the target table into the “MultiLoad” state. Upon successful completion, the target table is returned to the “normal” (non“MultiLoad”) state. Therefore, when a MultiLoad fails for any reason, the table is left in the “MultiLoad” state, and you cannot simply rerun the same MultiLoad. MultiLoad will report an error. In addition, MultiLoad also queries the target table’s MultiLoad log table to see if it contains any errors. If a MultiLoad log table exists for the target table, then you also will not be able to rerun your MultiLoad job. To recover from a failed MultiLoad, “release” the target table from the “MultiLoad” state and also drop the MultiLoad log table. You can do this using BTEQ or QueryMan to issue the following commands: drop table mldlog_<table name>; release mload <table name>; Note: The “drop table” command assumes that you’re recovering from a MultiLoad script generated by PowerCenter (PowerCenter always names the MultiLoad log table “mldlog_<table name>”). If you’re working with a hand-coded MultiLoad script, the name of the MultiLoad log table could be anything. Here is the actual text from a BTEQ session that cleans up a failed load to the table “td_test” owned by the user “infatest:” BTEQ -- Enter your DBC/SQL request or BTEQ command: drop table infatest.mldlog_td_test; drop table infatest.mldlog_td_test; *** Table has been dropped. *** Total elapsed time was 1 second. BTEQ -- Enter your DBC/SQL request or BTEQ command: release mload infatest.td_test; release mload infatest.td_test; *** Mload has been released. *** Total elapsed time was 1 second. Using One Instance of MultiLoad to Load Multiple Tables MultiLoad is a big consumer of resources on a Teradata system. Some systems will have hard limits on the number of concurrent MultiLoad sessions allowed. By default, PowerCenter will start an instance of MultiLoad for every target file. Sometimes, this is illegal (if the multiple instances target the same table). Other times, it is just expensive. Therefore, a prospect may ask that PowerCenter use a single instance of MultiLoad to load multiple tables (or to load both inserts and updates into the same target table). To make this happen, we must heavily edit the generated MultiLoad script file. Note: This should not be an issue with TPump because TPump is not as resource intensive as MultiLoad (and a multiple concurrent instances of TPump can target the same table). Here’s the workaround: 1) 2) 3) Use a dummy session (that is, set test rows to 1 and target a test database) to generate MultiLoad control files for each of the targets. Merge the multiple control files (one per target table) into a single control file (one for all target tables). Configure the session to call MultiLoad from a postsession script using the control file created in step 2. Integrated support cannot be used because each input file is processed sequentially and this causes problems when combined with the integrated named pipes and streaming of PowerCenter.
Details on “merging” the control files: 1) There is a single log file for each instance of MultiLoad. Therefore, you do not have to change or add anything the “LOGFILE” statement. However, you might want to change the name of the log table because it may be a log that spans multiple tables.
15 Informatica Confidential. Do not duplicate.
2) 3) 4) 5) 6) 7) 8)
Copy the work and error tables’ delete statements into the common control file. Modify the “BEGIN MLOAD” statement to specify all the tables that the MultiLoad will be hitting. Copy the “Layout” sections into the common control file and give each a unique name. Organize the file such that all the layout sections are grouped together. Copy the “DML” sections into the common control file and give each a unique name. Organize the file such that all the DML sections are grouped together. Copy the “Import” statements into the common control file and modify them to reflect the unique names created for the referenced layout and DML sections created in steps 4 and 5. Organize the file such that all the import sections are grouped together. Run “chmod –w” on the newly minted control file so PowerCenter doesn’t overwrite it, or, better yet, name it something different so PowerCenter cannot overwrite it. Remember, a single instance of MultiLoad can target at most five tables. Therefore, don’t combine more than five target files into a common file.
Here’s an example of a control file merged from two default control files: .DATEFORM ANSIDATE; .LOGON demo1099/infatest,infatest; .LOGTABLE infatest.mldlog_TD_TEST; DROP TABLE infatest.UV_TD_TEST ; DROP TABLE infatest.WT_TD_TEST ; DROP TABLE infatest.ET_TD_TEST ; DROP TABLE infatest.UV_TD_CUSTOMERS ; DROP TABLE infatest.WT_TD_CUSTOMERS ; DROP TABLE infatest.ET_TD_CUSTOMERS ; .ROUTE MESSAGES WITH ECHO TO FILE c:\LOGS\TgtFiles\td_test.out.ldrlog ; .BEGIN IMPORT MLOAD TABLES infatest.TD_TEST, infatest.TD_CUSTOMERS ERRLIMIT 1 CHECKPOINT 10000 TENACITY 10000 SESSIONS 1 SLEEP 6 ; /* Begin Layout Section */ .Layout InputFileLayout1; .Field .Field .Field .Field .Field .Field CUST_KEY 1 CHAR( 12) NULLIF CUST_KEY = '*' ; CUST_NAME 13 CHAR( 20) NULLIF CUST_NAME = '*' ; CUST_DATE 33 CHAR( 10) NULLIF CUST_DATE = '*' ; CUST_DATEmm 33 CHAR( 2) ; CUST_DATEdd 36 CHAR( 2) ; CUST_DATEyyyy 39 CHAR( 4) ;
16 Informatica Confidential. Do not duplicate.
.Field CUST_DATEtd CUST_DATE .Filler EOL_PAD
CUST_DATEyyyy||'/'||CUST_DATEmm||'/'||CUST_DATEdd NULLIF = '*' ; 43 CHAR( 2) ;
.Layout InputFileLayout2; .Field CUSTOMER_KEY 1 CHAR( 12) ; .Field CUSTOMER_ID 13 CHAR( 12) ; .Field COMPANY 25 CHAR( 50) NULLIF COMPANY = '*' ; .Field FIRST_NAME 75 CHAR( 30) NULLIF FIRST_NAME = '*' ; .Field LAST_NAME 105 CHAR( 30) NULLIF LAST_NAME = '*' ; .Field ADDRESS1 135 CHAR( 72) NULLIF ADDRESS1 = '*' ; .Field ADDRESS2 207 CHAR( 72) NULLIF ADDRESS2 = '*' ; .Field CITY 279 CHAR( 30) NULLIF CITY = '*' ; .Field STATE 309 CHAR( 2) NULLIF STATE = '*' ; .Field POSTAL_CODE 311 CHAR( 10) NULLIF POSTAL_CODE = '*' ; .Field PHONE 321 CHAR( 30) NULLIF PHONE = '*' ; .Field EMAIL 351 CHAR( 30) NULLIF EMAIL = '*' ; .Field REC_STATUS 381 CHAR( 1) NULLIF REC_STATUS = '*' ; .Filler EOL_PAD 382 CHAR( 2) ; /* End Layout Section */ /* begin DML Section */ .DML Label tagDML1; INSERT INTO infatest.TD_TEST ( CUST_KEY , CUST_NAME , CUST_DATE ) VALUES ( :CUST_KEY , :CUST_NAME , :CUST_DATEtd ) ; .DML Label tagDML2; INSERT INTO infatest.TD_CUSTOMERS ( CUSTOMER_KEY , CUSTOMER_ID , COMPANY , FIRST_NAME , LAST_NAME , ADDRESS1 , ADDRESS2 , CITY , STATE , POSTAL_CODE , PHONE , EMAIL , REC_STATUS ) VALUES (
17 Informatica Confidential. Do not duplicate.
:CUSTOMER_KEY :CUSTOMER_ID :COMPANY :FIRST_NAME :LAST_NAME :ADDRESS1 :ADDRESS2 :CITY :STATE :POSTAL_CODE :PHONE :EMAIL :REC_STATUS ) ; /* end DML Section */
,
, ,
, , , ,
, , , , ,
/* Begin Import Section */ .Import Infile c:\LOGS\TgtFiles\td_test.out Layout InputFileLayout1 Format Unformat Apply tagDML1 ; .Import Infile c:\LOGS\TgtFiles\td_customers.out Layout InputFileLayout2 Format Unformat Apply tagDML2 ; /* End Import Section */ .END MLOAD; .LOGOFF;
18 Informatica Confidential. Do not duplicate.
Multiple Workflows That MultiLoad to the Same Table Because MultiLoad puts a lock on the table, we require that all MultiLoad sessions must handle wait events so they don't try to access the table simultaneously. Also, any log files should be given unique names for the same reason. FastLoad As the name suggests, this utility is the fastest method to load data into Teradata. However, there is one major restriction: the target table must be empty.
19 Informatica Confidential. Do not duplicate.
Teradata Warehouse Builder (TWB) Teradata Warehouse Builder (TWB) is a single utility that was intended to replace FastLoad, MultiLoad, TPump, and FastExport. It was to support a single scripting environment with different “modes,” where each “mode” roughly equates to one of the legacy utilities. It also was to support parallel loading (that is, multiple instances of a TWB client could run and load the same table at the same time—something the legacy loaders cannot do). Although PowerCenter supports TWB, NCR/Teradata does not. TWB has never been formally released. According to NCR, its general release was delayed primarily because of issues with the mainframe version. If you find a prospect willing to use TWB, please do. Its ability to support parallel load clients makes some tasks easier.
20 Informatica Confidential. Do not duplicate.
PowerCenter 7.1.2 PAM for Teradata Server Client Software Platform Version & Version Source Target
Comment
PowerCenter uses the following components of the Teradata Tools and Utilities (TTU): Teradata ODBC driver, FastLoad, MultiLoad, and TPump. The version numbers of each of the Teradata Client components vary with release (e.g., TTU7 contains ODBC 3.02, FastLoad 7.05, MultiLoad 3.03, TPump 1.07). The TTU was previously called Teradata Utilities Foundation (TUF). Compatibility between a particular version of the Teradata RDBMS and the Teradata Client software is determined by Teradata -- not Informatica.
NCR Teradata
UNIX and NT
V2R4
TUF 6.1, TTU 6.1.1, TTU 6.2
x
x
UNIX and NT
TUF 6.1, TTU 6.1.1, V2R4.1 TTU 6.2, TTU 7
x
x The Teradata Client/Teradata RDBMS pairings listed here represent our understanding based on Teradata's documentation. Note that the minimum version number for the Teradata ODBC driver is 3.00.01.04. Teradata has made many fixes to the 3.02 ODBC driver. If you are using this driver, please contact NCR support for the latest maintenance release.
UNIX and NT
V2R5
TTU 7
x
x
UNIX and NT
V2R5.1 TTU 7
x
x
UNIX and NT
V2R6
TTU 8
x
x
Supported Teradata ODBC driver is v3.0.4.1.
21 Informatica Confidential. Do not duplicate.
New PowerCenter Features Not Covered in this Document
FastExport As its name implies, FastExport is a utility to extract data from Teradata very quickly. It is supported in PowerCenter 7.1.3 for Teradata sources. This feature will extract large tables used for lookups quicker than ODBC,. Pushdown Optimization The upcoming release of PowerCenter 8 will be feature a Pushdown Optimization Option. This option enables data transformation processing, where appropriate, to be “pushed down” into a relational database. Powered by an easy-to-use, visual user interface, the option streamlines the generation and execution of platform-specific business logic and manages the execution in a database as opposed to the Informatica Data Server. The option is particularly useful in scenarios where data is “co-resident” (i.e., staging and production are in the same database instance). With this new functionality, users will be able to create SQL that will execute in the Teradata database server, replacing certain transformations that would normally run in the Informatica Data Server. This is critical to Teradata because the Source and Target commonly reside in the same database.
©Copyright 2005 Informatica Corporation. Informatica and PowerCenter are registered trademarks of Informatica Corporation. Teradata is a registered trademark of NCR Corporation. All other company, product, or service names may be trademarks or registered trademarks of their respective owners.
J50665 (9/29/2005)
22 Informatica Confidential. Do not duplicate.
