Teradata Components

Teradata Components

There are three main components in Teradata which does all wonder to the world of Database.

1. Parsing Engine

1. AMP (Access Module Processor)

1. Bynet 

Parsing Engine (PE)

The Teradata Parsing Engine (PE) is the hero in the world of database optimization. Most databases make educated guesses about the best way to retrieve data. Teradata has such a great reputation and experience for speeding up data access that it has earned the name “The OPTIMIZER.” Because Teradata knows how to retrieve data quickly it gives users the confidence to believe in the data warehouse. As users gain experience they begin to believe in themselves.

When you logon to Teradata, they connect to a Parsing Engine (PE). When a user submits a query, then the PE takes action. The PE creates a PLAN that tells the AMPs exactly what to do in order to get the data. The PE knows how many AMPs are in the system, how many rows are in the table, and the best way to get to the data. The PE enjoys serving valid Teradata users, but it was raised like a guard dog. A good guard dog loves its family, but it barks and may bite when strangers approach. The PE will always check a user’s security (access) rights to ensure they have the proper authority to obtain the information that is being requested. If the user has authority, the PE instructs the AMPs to get the data. If the user doesn’t have proper access rights, the query is rejected.

The PE doesn’t like to brag, but it did graduate at the top of its class. This has given the PE years of experience in guiding the AMPs to answer complex questions – some of which have never been asked before in their respective industries. This experience allows users to ask any question regardless of its complexity. The PE isn’t called “The Optimizer” for nothing. It needs no tuning by a Database Administrator (DBA) or hints from the user. Teradata users ask the questions, and Teradata simply returns the answers.

Access Module Processor (AMP)

The Access Module Processor (AMP) is a component of little words. It keeps its mouth shut and its ears open. Each AMP listens to the PE via the BYNET for instructions. The AMPs primary responsibilities consist of retrieving and writing data to its disk. The AMP is the worker bee of the system. It is the perfect employee. It never complains, rarely calls in sick, and lives to take direction from its boss who is the Parsing Engine (PE). The AMPs work together like many hands to make the work light.

Every AMP has its own disk, and it’s the only AMP allowed to read or write data to that disk. This method of processing is referred to as a “Shared-Nothing” architecture. Although AMPs are the perfect workers, they are not the perfect playmates. Even as children AMPs would never share toys with other AMPs on the playground. Each AMP has its own disk, and it shares this with no other AMP, hence a “Shared-Nothing” architecture.

Teradata spreads the rows of a table evenly across all AMPs in the system. When the PE asks the AMPs to get the data, each AMP will read the rows only on their particular disk. If this is done simultaneously, all AMPs should finish at about the same time. As a matter of fact, when we explained this philosophy to Confucius he stated, “A query is only as fast as the slowest AMP.” Confucius, however, did say not to quote him!

Again, an AMPs job is to read and write data to its disk. The AMP takes its direction from the Parsing Engine (PE). The number of AMPs varies per system. 

The BYNET

The BYNET ensures communication between AMPs and PEs. When fast communication is necessary, the BYNET operates as a communications superhighway. There are no traffic jams on the BYNET because its designers went the extra mile. The BYNET makes sure that communication between AMPs and PEs are always on the right track and that it happens rapidly. The BYNET has been designed for speed, reliability, and throughput.

There are actually two BYNETs per system. They are called “BYNET 0” and “BYNET 1”. The two BYNETs are usually referred to as just the BYNET. The reasons two BYNETs exist on a Teradata system is because of the following:

1. Redundancy: If one BYNET fails, the second BYNET takes over.

2. Performance: Think of the two BYNETs as telephone lines in your home. AMPs and PE’s can talk to one another on one or over both BYNETs.

Here are the steps that outline exactly how the AMPs, PEs, and BYNET work together: A user performs a LOGON to Teradata. A PE is assigned to manage all SQL requests for that particular user. When a user queries Teradata, the following occurs: The PE checks the user’s SQL Syntax;

·       The PE checks the user’s security rights;

·       The PE comes up with a plan for the AMPs to follow;

·       The PE passes the plan along to the AMPs over the BYNET;

·       The AMPs follow the plan and retrieve the data requested;

·       The AMPs pass the data to the PE over the BYNET; and

·       The PE then passes the final data to the user.

The BYNET provides the communications between AMPs and PEs – so no matter how large the data warehouse physically gets, the BYNET makes each AMP and PE think that they are right next to one another. The BYNET gets its name from the Banyan tree. The Banyan tree has the ability to continually plant new roots to grow forever. In addition, Banyan trees have been found that are a mile wide. The BANYON network or BYNET has the ability to continually grow. The BYNET is installed at the largest data warehouse sites in the world and the DBA’s smiles that are a mile wide.