Teradata

Understanding Teradata: The Enterprise Data Warehouse Powerhouse

In the era of big data, where businesses generate and rely on vast amounts of information to make critical decisions, having an efficient, scalable, and reliable data warehouse system is essential. One of the leading solutions in this space is Teradata, a high-performance relational database management system (RDBMS) designed specifically for large-scale data warehousing and analytics. Teradata has earned a reputation for its ability to handle massive volumes of structured and semi-structured data, making it a preferred choice for enterprises across industries like finance, retail, telecommunications, and healthcare.

History and Evolution of Teradata

Teradata’s origins trace back to the 1970s when researchers at the University of California, Irvine, began developing parallel processing database systems. However, the commercial entity Teradata Corporation was officially founded in 1979 and later became part of NCR Corporation in the 1990s. Teradata distinguished itself by pioneering massively parallel processing (MPP) technology for relational databases, which allowed it to scale horizontally and handle extremely large datasets more efficiently than traditional RDBMSs like Oracle or SQL Server.

Over the years, Teradata has evolved beyond an on-premises data warehouse solution to include cloud-based offerings such as Teradata Vantage, which integrates AI, machine learning, and multi-cloud capabilities, enabling modern enterprises to perform analytics at scale across hybrid and cloud environments.

Core Architecture of Teradata

Teradata’s strength lies in its shared-nothing architecture combined with massively parallel processing (MPP). These design principles ensure high scalability, fault tolerance, and fast query performance.

1. Shared-Nothing Architecture:
   In a shared-nothing architecture, each node in the Teradata system has its own memory and disk storage. Unlike shared-disk systems, there is no contention for resources between nodes, allowing multiple queries to execute concurrently without interference. This design ensures that as data grows, the system can scale linearly by simply adding more nodes.

2. Massively Parallel Processing (MPP):
   Teradata distributes data evenly across multiple processing units called AMPs (Access Module Processors). Each AMP is responsible for a portion of the data, enabling parallel execution of queries. When a SQL query is executed, Teradata breaks it down into smaller tasks, distributes them across all AMPs, and aggregates the results. This parallelization significantly reduces query response times for large datasets.

3. Teradata File System and Data Distribution:
   Teradata uses a hashing algorithm to distribute rows across AMPs evenly. This hashing ensures that data is balanced and that no single AMP becomes a bottleneck. Combined with its indexing strategies, Teradata achieves high-speed access and retrieval, even for tables containing billions of rows.

4. Fallback and Recovery Mechanisms:
   Teradata ensures reliability with fallback mechanisms that replicate data across multiple nodes. In case of node failure, the system automatically retrieves the data from the fallback copy, ensuring no downtime and high availability.

Teradata Vantage: The Modern Analytics Platform

In recent years, Teradata has transitioned from a traditional data warehouse provider to a comprehensive analytics platform. Teradata Vantage is the company’s flagship offering that combines data warehousing, analytics, and machine learning capabilities in a unified platform.

Key features of Teradata Vantage include:

• Multi-Cloud and Hybrid Deployment: Vantage can run on public clouds like AWS, Azure, Google Cloud, or on-premises, enabling flexibility for enterprises.

• Integrated Analytics: Vantage allows data scientists and analysts to run machine learning, graph, and geospatial analytics directly within the database, minimizing the need to move data between systems.

• High-Performance Query Engine: The MPP engine, combined with advanced indexing and query optimization, enables near real-time insights from petabytes of data.

• Support for Modern Data Formats: In addition to traditional structured data, Vantage can handle semi-structured data like JSON, Avro, and Parquet, making it suitable for modern analytics workloads.

Teradata vs. Other Data Warehouses

Teradata has often been compared with other enterprise data warehouse solutions like Snowflake, Oracle Exadata, and Microsoft SQL Server. Here’s how Teradata stands out:

1. Scalability: Teradata’s shared-nothing, MPP architecture allows linear scaling, meaning performance increases proportionally as nodes are added.

2. Query Performance: Its ability to parallelize queries across AMPs allows it to outperform many traditional relational databases for large datasets.

3. Enterprise-Grade Reliability: Teradata is designed for high availability, fault tolerance, and data integrity, which is critical for large organizations handling mission-critical workloads.

4. Advanced Analytics Integration: Unlike some legacy systems, Teradata integrates advanced analytics, AI, and machine learning capabilities directly within the platform.

While newer cloud-native solutions like Snowflake excel in flexibility and cost efficiency, Teradata remains a preferred choice for enterprises requiring predictable, high-speed performance for complex, large-scale analytics workloads.

Real-World Applications of Teradata

Teradata is widely used in industries where large-scale analytics drive business decisions:

• Retail: Retail giants use Teradata to analyze customer behavior, optimize inventory, and drive personalized marketing campaigns.

• Finance: Banks and insurance companies rely on Teradata for risk analysis, fraud detection, and regulatory reporting.

• Telecommunications: Telecom operators leverage Teradata for customer churn analysis, network optimization, and targeted promotions.

• Healthcare: Hospitals and pharmaceutical companies use Teradata to integrate and analyze clinical, operational, and patient data for better outcomes.

Challenges and Considerations

Despite its strengths, Teradata has some challenges:

• Cost: Traditional Teradata implementations can be expensive, particularly on-premises setups.

• Complexity: Setting up and maintaining Teradata systems requires specialized skills.

• Cloud Transition: While Teradata Vantage supports cloud deployments, some organizations find it challenging to migrate legacy on-premises workloads.

However, the company has made significant strides with its cloud-first approach, offering flexible consumption models and managed services to address these challenges.

Conclusion

Teradata remains a cornerstone in the world of enterprise data warehousing and analytics. Its unique architecture, proven reliability, and advanced analytics capabilities make it suitable for organizations managing massive amounts of data and requiring high-performance insights. As businesses continue to embrace data-driven strategies, platforms like Teradata Vantage demonstrate how traditional data warehouse solutions can evolve to meet modern analytics needs, bridging the gap between structured enterprise data and cutting-edge AI and machine learning applications.

For enterprises seeking a robust, scalable, and enterprise-grade data platform, Teradata continues to offer unmatched performance, reliability, and integration capabilities that enable smarter, faster, and more strategic decision-making.

1. Basics of Teradata

Q1. What is Teradata?
Answer:
Teradata is a Relational Database Management System (RDBMS) designed for large-scale data warehousing applications. It uses Massively Parallel Processing (MPP) to handle huge volumes of data efficiently. Teradata is widely used in analytics, reporting, and business intelligence.

Key Points:

• Handles petabytes of data.

• Uses Parallelism: AMPs (Access Module Processors) distribute data.

• SQL compliant with extensions for analytic functions.

Q2. What are the features of Teradata?
Answer:

• Parallel Processing – AMPs process data concurrently.

• Scalability – Can scale horizontally by adding nodes.

• High Availability – Fault tolerance and data protection.

• Optimized for SQL queries – Has Teradata Optimizer.

• Support for large data sets – Handles structured and semi-structured data.

Q3. What are the components of Teradata Architecture?
Answer:

1. Parsing Engine (PE): Parses SQL queries, checks syntax, and creates execution plans.

2. AMP (Access Module Processor): Performs actual data retrieval and storage operations.

3. BYNET: Interconnect network connecting nodes for data communication.

4. Teradata Database (Data Storage): Stores user data in tables across AMPs.

Q4. What is an AMP?
Answer:
AMP (Access Module Processor) is a virtual processor responsible for storing and retrieving a portion of the data. Each AMP has its own disk storage and operates in parallel.

Example: If you have 4 AMPs and 1,000 rows, each AMP might hold ~250 rows.

Q5. Explain Primary Index in Teradata.
Answer:
A Primary Index (PI) determines how data is distributed across AMPs.

• Unique PI (UPI): Ensures unique values; one row per AMP location.

• Non-Unique PI (NUPI): Can have duplicates; multiple rows can go to the same AMP.

Example:

CREATE TABLE Employee(
   EmpID INT,
   EmpName VARCHAR(50),
   DeptID INT
) PRIMARY INDEX(EmpID);

Here, EmpID decides how rows are hashed across AMPs.

Q6. What is a Primary Key vs Primary Index?
Answer:

• Primary Key (PK): Ensures uniqueness of a row.

• Primary Index (PI): Determines physical storage location on AMPs.

PK may or may not be PI.

Q7. What is Secondary Index (SI) in Teradata?
Answer:
Secondary Index allows alternate access paths to data without changing the primary distribution.

• Unique SI (USI): Each value points to exactly one row.

• Non-Unique SI (NUSI): Value can point to multiple rows.

Helps improve query performance but adds storage overhead.

Q8. What is a Join Index?
Answer:
A Join Index stores pre-joined data to improve query performance.

• Types: Single-table Join Index, Multi-table Join Index.

• Reduces join computation at query runtime.

Q9. What are Teradata Data Types?
Answer:

• Numeric: BYTEINT, SMALLINT, INTEGER, BIGINT, DECIMAL, NUMERIC.

• Character: CHAR, VARCHAR, CLOB.

• Date/Time: DATE, TIME, TIMESTAMP, INTERVAL.

• LOBs: BLOB, CLOB.

Q10. What is Teradata SQL vs ANSI SQL?
Answer:

• Teradata SQL: Supports Teradata-specific functions (e.g., QUALIFY, SAMPLE).

• ANSI SQL: Standard SQL; portable across RDBMS.

• Teradata supports both, but Teradata SQL has better performance in Teradata environment.

2. Teradata Table Concepts

Q11. What are the types of tables in Teradata?
Answer:

1. Permanent Table: Stored permanently; data persists.

2. Volatile Table: Exists for the session only; no logging to disk.

3. Global Temporary Table (GTT): Data session-specific; structure permanent.

4. Derived Table: Temporary table created via SELECT.

Q12. What is a Multi-Set Table?
Answer:

• Allows duplicate rows.

• Default in Teradata.

Example:

CREATE MULTISET TABLE Employee(...);

Q13. What is a Set Table?
Answer:

• Does not allow duplicate rows.

• Unique rows enforced by Teradata.

Q14. What is Normalization and Denormalization?
Answer:

• Normalization: Organize data to reduce redundancy.

• Denormalization: Merge tables to improve query performance.

Teradata often uses denormalized tables for reporting because fewer joins improve performance.

Q15. What is a Sparse Table in Teradata?
Answer:

• Table that doesn’t allocate storage for null columns.

• Improves storage efficiency when many nulls exist.

Q16. Explain Partitioned Primary Index (PPI).
Answer:

• PPI partitions data within each AMP based on column values.

• Improves query performance for range-based queries.

Example:

CREATE TABLE Sales(
   SaleID INT,
   SaleDate DATE,
   Amount DECIMAL(10,2)
) PRIMARY INDEX(SaleID)
PARTITION BY RANGE_N(SaleDate BETWEEN DATE '2026-01-01' AND DATE '2026-12-31' EACH INTERVAL '1' MONTH);

3. Teradata Performance & Query Concepts

Q17. What is Teradata Optimizer?
Answer:

• Teradata Optimizer decides efficient execution plans for SQL queries.

• It considers:

  • Data distribution (PI)

  • Table statistics

  • Join strategy (product join, merge join, etc.)

Q18. What are the types of Joins in Teradata?
Answer:

1. Merge Join: Sorted tables, best for large tables.

2. Product Join (Cartesian): Each row with each row; expensive.

3. Hash Join: Uses hashing, common in Teradata.

Q19. What is a Collect Statistics in Teradata?
Answer:

• COLLECT STATISTICS gathers metadata about column data distribution.

• Helps optimizer choose the best query plan.

Example:

COLLECT STATISTICS ON Employee COLUMN EmpID;

Q20. What is Skew in Teradata?
Answer:

• Skew happens when data is unevenly distributed across AMPs.

• Can slow queries and reduce parallelism.

Solution: Choose good primary index to avoid skew.

Q21. How to Avoid Skew in Teradata?
Answer:

• Choose columns with high cardinality as PI.

• Avoid NUPI on low-cardinality columns.

Q22. Explain Teradata Temporary Tables
Answer:

• Volatile Table: Exists only in session; no permanent storage.

• Global Temporary Table: Structure permanent, data session-specific.

• Derived Table: Created in SQL query for intermediate results.

Q23. What is a Teradata Macro?
Answer:

• Predefined set of SQL statements stored in Teradata.

• Helps reuse queries.

CREATE MACRO emp_macro AS
(SELECT EmpName, DeptID FROM Employee);

Q24. What is Teradata Stored Procedure?
Answer:

• Block of SQL statements executed together.

• Supports variables, control flow, loops.

Q25. Explain Teradata Views
Answer:

• Logical representation of a table or join.

• Types:

  • Simple View: Single table.

  • Complex View: Join multiple tables, use aggregation.

Q26. What is Teradata BTEQ?
Answer:

• Basic Teradata Query tool for running queries, exporting, and formatting reports.

• Modes: Interactive and Batch.

Q27. Explain FastLoad, MultiLoad, and TPump
Answer:

• FastLoad: Loads empty tables fast.

• MultiLoad: Load multiple tables, supports updates and deletes.

• TPump: Load continuously with low latency.

Q28. What is Teradata Fallback?
Answer:

• Creates duplicate copy of a row on another AMP for high availability.

• Ensures data recovery in case of AMP failure.

Q29. What is Teradata Hashing Algorithm?
Answer:

• Hash function maps primary index value to AMP.

• Ensures even distribution and parallel processing.

Q30. What are Teradata Operators?
Answer:

• Join, Union, Aggregate, Sort, Project, Restrict.

• Teradata executes queries via Relational Operators.

4. Practical SQL & Query Questions

Q31. How to get top N rows in Teradata?
Answer:

SELECT * FROM Employee
QUALIFY ROW_NUMBER() OVER (ORDER BY Salary DESC) <= 10;

Q32. How to delete duplicate rows in Teradata?
Answer:

DELETE FROM Employee
WHERE EmpID NOT IN (
  SELECT MIN(EmpID)
  FROM Employee
  GROUP BY EmpName
);

Q33. Difference between DELETE and TRUNCATE?
Answer:

• DELETE: Row-level removal; can have WHERE clause; logs transactions.

• TRUNCATE: Table-level removal; faster; no WHERE clause; resets identity.

Q34. Explain CASE and COALESCE in Teradata
Answer:

• CASE: Conditional expression.

• COALESCE: Returns first non-null value.

Q35. What is SAMPLE in Teradata?
Answer:

• Used to fetch random rows.

SELECT * FROM Employee SAMPLE 10;

Q36. Difference between QUALIFY and WHERE
Answer:

• WHERE: Filters before aggregation.

• QUALIFY: Filters after window functions like ROW_NUMBER().

Q37. What are Teradata Set Operators?
Answer:

• UNION, INTERSECT, EXCEPT – similar to ANSI SQL.

Q38. Explain Teradata Aggregation Functions
Answer:

• SUM, COUNT, AVG, MIN, MAX

• Aggregate over groups with GROUP BY.

Q39. Explain Teradata OLTP vs OLAP
Answer:

• OLTP: Transactional; inserts/updates/deletes; row-level operations.

• OLAP: Analytical; data warehouse; query-intensive.

Q40. What is Teradata Export and Import?
Answer:

• Export: Export data using BTEQ, FastExport.

• Import: Load data using FastLoad, MultiLoad, TPump.

Q41. Common Teradata Interview Tips for Freshers:

• Understand primary index and data distribution.

• Learn basic SQL and joins thoroughly.

• Be able to explain Volatile vs Permanent tables.

• Know BTEQ, FastLoad, MultiLoad basics.

• Read EXPLAIN plans to interpret query execution.

1. Teradata Basics & Architecture

Q1. What is Teradata and what are its main features?
Answer:
Teradata is a massively parallel processing (MPP) relational database management system (RDBMS) designed for large-scale data warehousing. Key features:

• Shared Nothing Architecture: Each AMP (Access Module Processor) has its own CPU, memory, and disk.

• Massively Parallel Processing (MPP): Parallel execution of queries improves performance.

• Scalability: Can scale horizontally by adding more nodes.

• Teradata Optimizer: Advanced query optimizer for efficient query execution.

• High Availability & Fault Tolerance: Automatic failover and data recovery.

Q2. Explain Teradata Architecture.
Answer:
Teradata has a 3-layer architecture:

1. Parsing Engine (PE): Receives SQL, parses it, checks syntax & semantics, and creates an execution plan.

2. Message Passing Layer (BYNET): Network layer that connects PEs and AMPs for message passing.

3. Access Module Processors (AMPs): Data storage and retrieval. AMPs execute the plan in parallel.

Key points: PE → BYNET → AMP, then results are returned via BYNET to PE.

Q3. What is an AMP?
Answer:
AMP (Access Module Processor) is a virtual processor responsible for:

• Storing a portion of the data (rows)

• Performing database operations like insert, update, delete

• Each AMP has its own disk space called PDE (Physical Data Engine)

Data distribution: Teradata distributes rows using a hash function on Primary Index to ensure even distribution across AMPs.

Q4. Explain Primary Index and its types.
Answer:
Primary Index determines how data is distributed across AMPs. Two types:

1. Unique Primary Index (UPI): Ensures each row is unique. Good for evenly distributing data.

2. Non-Unique Primary Index (NUPI): Rows may have duplicates. Can cause skew if many duplicates exist.

Important: Choosing the right PI is crucial for performance.

Q5. What is Primary Key vs Primary Index in Teradata?
Answer:

• Primary Key (PK): Ensures uniqueness logically, no impact on physical data distribution.

• Primary Index (PI): Determines row distribution across AMPs. Can be unique or non-unique.

Tip: PK can be non-PI, PI can be non-unique.

2. Teradata SQL & Query Optimization

Q6. Explain Teradata SQL differences vs standard SQL.
Answer:

• Teradata supports ANSI SQL but has some unique features:

  • QUALIFY to filter after analytic functions

  • SAMPLE for random sampling

  • TOP vs LIMIT syntax

  • SPOOL space management

Q7. What is SPOOL space in Teradata?
Answer:

• Temporary space used to store intermediate query results

• Stored on AMPs

• Important: Queries fail if they exceed SPOOL limits

Q8. What is a Join in Teradata and types of joins?
Answer:
Teradata supports:

• Inner Join: Returns matching rows from both tables

• Left/Right Outer Join: Returns all rows from one table + matching rows from other

• Full Outer Join: All rows from both tables

• Product Join (Cross Join): Cartesian product

• Hash Join & Merge Join: Optimized joins using AMP-local or sorted data

Q9. Explain Teradata’s Hash Join mechanism.
Answer:

• Hash join uses Primary Index hash values.

• Matching rows are joined on the same AMP.

• Avoids data redistribution (no movement across AMPs).

Q10. What is a Cross-AMP vs Local Join?
Answer:

• Local Join: Data exists on the same AMP; no data redistribution needed.

• Cross-AMP Join: Data needs redistribution to AMPs, which is expensive (affects performance).

Optimization tip: Always try to join on columns that are Primary Index to enable Local Joins.

Q11. What is a Teradata Volatile Table?
Answer:

• Temporary table exists only for the session

• Stored in SPOOL space

• No DBC privileges required

• Auto-dropped at session end

Example:

CREATE VOLATILE TABLE temp_sales AS 
(SELECT * FROM sales WHERE sale_date > DATE '2026-01-01') WITH DATA 
ON COMMIT PRESERVE ROWS;

Q12. What are Derived Tables and CTEs?
Answer:

• Derived Table: Subquery in FROM clause. Temporary and used only once.

• CTE (Common Table Expression): Named temporary result set defined by WITH clause; can be reused.

Q13. Explain the QUALIFY clause in Teradata.
Answer:

• Filters rows after window/analytic functions

• Example: Find top 1 sale per region:

SELECT region, sale, RANK() OVER(PARTITION BY region ORDER BY sale DESC) AS rnk
FROM sales
QUALIFY rnk = 1;

Q14. What is an Aggregate Join?
Answer:

• Aggregate first, then join

• Reduces data movement across AMPs

• Improves query performance

Q15. How do you identify query bottlenecks in Teradata?
Answer:

• Use EXPLAIN statement to analyze execution plan

• Check for:

  • Cross-AMP redistributions

  • Skewed AMPs

  • Large SPOOL usage

  • Cartesian products

3. Data Distribution & Performance Tuning

Q16. How do you avoid data skew in Teradata?
Answer:

• Choose a Primary Index with high cardinality

• Avoid using low-cardinality columns

• Use hash-distributed tables for large datasets

• Monitor with HELP STATISTICS

Q17. What is a Secondary Index (SI)?
Answer:

• Additional access path for rows without affecting PI

• Unique SI: 1 row per value, uses a separate table

• Non-Unique SI: Multiple rows, can reduce full table scans

Q18. Explain Join Index in Teradata.
Answer:

• Pre-joined table for performance

• Can be unique/non-unique

• Helps to speed up frequent complex joins

Q19. Explain Hashing in Teradata.
Answer:

• PI columns are hashed using Teradata’s hash algorithm

• Determines AMP allocation

• Ensures even data distribution

Q20. What is Partitioned Primary Index (PPI)?
Answer:

• PI column + partitioning column

• Improves range queries

• Reduces full AMP scans

• Example: Partition by date ranges

Q21. What is Teradata’s Multi-Value Compression (MVC)?
Answer:

• Compress frequently occurring column values

• Reduces storage

• Example: status column with repeated 'Active' values

Q22. What is Teradata’s FastLoad?
Answer:

• Utility to load large tables fast

• Only for empty target tables

• Loads data in parallel

• Cannot update existing tables

Q23. What is MultiLoad?
Answer:

• Utility to load multiple tables

• Supports insert, update, delete, upsert

• Suitable for incremental loads

Q24. What is TPump?
Answer:

• Utility for continuous loading

• Handles small batches efficiently

• Maintains transaction logging

Q25. What is BTEQ?
Answer:

• Command-line tool for running SQL scripts

• Supports export, import, reporting

• Can be interactive or batch mode

Q26. How do you monitor Teradata system performance?
Answer:

• Use DBQL (Database Query Log) to track query performance

• DBC.ResUsage to monitor resource usage

• Check skew using SHOW TABLE and HELP STATS

Q27. Explain Teradata Statistics and why they are important.
Answer:

• Help optimizer choose the best execution plan

• Types: Column Stats, Multi-Column Stats, Index Stats

• Example:

COLLECT STATISTICS ON sales COLUMN (sale_date);

Q28. Explain Teradata Explain Plan.
Answer:

• EXPLAIN <SQL> shows execution steps, including:

  • Step number

  • Type of join

  • Data movement

  • AMP operations

Q29. What is an Ordered vs Unordered Primary Index?
Answer:

• UPI (Unique Primary Index): Default unordered; evenly distributed

• Ordered PI (OPI): Data physically sorted on disk; efficient for range queries

Q30. What is Teradata fallback?
Answer:

• Data redundancy for fault tolerance

• If AMP fails, copy on another AMP is used

• Can be enabled or disabled per table

4. Advanced Topics

Q31. What are Macros and Stored Procedures in Teradata?
Answer:

• Macro: Predefined SQL block; no parameters

• Stored Procedure: Can accept parameters, support control flow logic (IF, LOOP)

Q32. What are UDFs (User Defined Functions)?
Answer:

• Custom functions for specific processing

• Can be scalar or aggregate functions

Q33. Explain Teradata’s Join Strategies
Answer:

• Merge Join: Sorts tables, merges on PI/Join column

• Product Join: Cartesian join, expensive

• Hash Join: Most efficient, uses hash values

Q34. Explain Teradata Temp Tables vs Global Temp Tables
Answer:

• Volatile Table: Session-specific, auto-drop

• Global Temporary Table: Defined once, exists across sessions, rows are session-specific

Q35. How do you optimize ETL in Teradata?
Answer:

• Use primary index wisely

• Avoid unnecessary joins

• Collect statistics before large ETL runs

• Use FastLoad/MultiLoad for bulk loading

Q36. How does Teradata handle NULL values in PI/NUPI?
Answer:

• UPI cannot have NULLs

• NUPI can have NULLs, but many NULLs can cause skew

Q37. What is Teradata Temporal Tables?
Answer:

• Stores historical data

• Two types:

  • Transaction Time: History of changes

  • Valid Time: Validity period for business use

Q38. How do you debug a slow Teradata query?
Answer:

1. Check EXPLAIN plan

2. Look for skewed AMPs

3. Check for large SPOOL usage

4. Optimize joins and filters

5. Collect/update statistics

Q39. What is Teradata Query Banding?
Answer:

• Attaches metadata to sessions for tracking queries

• Useful for auditing, workload management

Q40. What is Teradata Archive/Recovery?
Answer:

• Backup tables using ARC or TPump

• Restore using FastLoad or MultiLoad

• Fallback ensures AMP-level recovery

Q41. Explain Difference Between ANSI & Teradata Session Mode
Answer:

• ANSI Mode: Compliant with ANSI standards; allows transactions to rollback; stricter on NULLs

• Teradata Mode: Legacy mode; different treatment of NULLs; faster in some cases