TDVAN5 Exam Questions & Answers

IO SKEW refers to an uneven distribution of I/O operations across AMPs, which can occur during large redistributions or full table scans, as data needs to be read from or written to the entire table. High IO skew indicates that some AMPs are handling significantly more I/O than others, which is often a sign of inefficient data distribution or full table scanning.

QRY_B has an IO SKEW value of 2.76, which is much higher than the other queries, suggesting that it is likely performing a full table scan or large redistribution, as this would result in uneven I/O across AMPs.

CPU SKEW (as seen in QRY_D) could indicate an issue with CPU load balancing, but it doesn't necessarily suggest a full table scan or redistribution.

UII (Utility Impact Index) and PJI (Product Join Index) are also metrics that might indicate query inefficiencies, but the high IO SKEW in QRY_B is the most direct indicator of a full table scan or large redistribution.

QRY_B is the query suspected of causing the full table scan or large redistribution due to its high IO SKEW value.

If a table in the DBC (Data Dictionary) database is corrupted, the proper approach is to set the corrupted AMP offline to isolate the issue, and then restore the corrupted table from a backup. This limits the scope of the recovery to only the affected AMP and table, minimizing the impact on the rest of the system.

Perform a SYSINIT, and restore the corrupted table: A SYSINIT is a full system initialization and should only be done in extreme cases, as it wipes the entire system. This is too drastic for table-level corruption.

Set the corrupted AMP offline, and restore the entire system: Restoring the entire system is unnecessary if only one table is corrupted.

Perform a SYSINIT, and restore the entire system: This would be an extreme and time-consuming option, only used if the entire system is corrupted.

An Online backup strategy allows the database to remain accessible and operational while the backup is taking place. This means that users can continue to modify data, and the system can remain available without requiring downtime. It is a common approach for mission-critical systems that need to maximize uptime.

Offline would require the system or certain tables to be unavailable during the backup process, which is not suitable for a system that needs to remain available.

Skip statistics refers to skipping the backup of statistics on database objects and does not relate to whether the system can be updated during the backup.

Dictionary Only involves backing up only the system catalog or dictionary data, not the actual table data, and would not fulfill the requirement of a comprehensive backup while allowing updates.

Data Mover Server should be deployed in the source environment (production) for optimum data transfer. This setup ensures that the data is transferred as efficiently as possible from the production system to the disaster recovery (DR) system. By having the server close to the source, it can access the data more quickly and efficiently initiate the transfer process, minimizing delays caused by geographic distance.

Location does not affect data transfer bandwidth: Location does affect data transfer bandwidth due to network latency and distance between systems.

In the cloud: Using a cloud environment would introduce unnecessary complexity and potential latency since both systems are on-prem.

In the target environment (DR): Deploying the Data Mover Server in the target environment could introduce latency issues, as it would have to pull data from the production system over long distances.

Thus, placing the server in the source environment is optimal for reducing latency and maximizing data transfer efficiency.

In this scenario, the issue arises due to high concurrency of stored procedure calls, which are exhausting AWTs (AMP Worker Tasks). To address this, the administrator needs to throttle the specific stored procedure calls made by BATCH_USER. The most effective way to do this is to create a new workload that specifically targets those calls (statement type = CALL) from BATCH_USER and then applies a throttle to limit concurrency. This helps manage the system's AWTs by controlling how many stored procedures can be run concurrently.

Other options are less suitable as they either do not target stored procedures specifically or may not provide the necessary granularity in controlling AWT usage for stored procedure calls.