Data Skew Issue — Learn free

Self-check

Questions 1–10 of 10

1. What causes a Data Skew Issue?
Question 1A. Data not being distributed evenly across the nodes in a clusterB. Indexes being disabled on large tablesC. Poorly written join conditions in a queryD. Low data quality in the source systemsE. A lack of data standardisation across datasets
2. What is the Data Skew Issue normally referred to as in the Data Science domain?
Question 2A. Data CouplingB. Uneven Load BalancingC. Data EnrichmentD. Skewed Data DistributionE. Massive Parallel Processing
3. Which kind of query is said to suffer the most from a Data Skew Issue?
Question 3A. Simple single-table lookupsB. Queries with joinsC. Queries that only count rowsD. Queries run against a single nodeE. Queries on freshly loaded datasets
4. Why does uneven data distribution slow queries in a Massive Parallel Processing (MPP) cluster?
Question 4A. Because the cluster has to rebuild every index before the query runsB. Because skewed data forces the query to run on a single node onlyC. Because uneven data automatically triggers a full backupD. Because MPP switches to a non-parallel mode when data is unequalE. Because MPP cannot be utilised properly, so the compute power of all nodes is not fully used
5. Which activity is named as one of the main reasons for a Data Skew Issue?
Question 5A. Adding more CPU and RAM to the clusterB. Enabling indexes on frequently joined columnsC. Data Enrichment, such as inserts, updates and delete tags that move data between locationsD. Running queries during peak business hoursE. Applying data standardisation rules
6. In the ten-node cluster example for data skew, three nodes end up with 2% of the data each while three others gain an extra 8% each. What happens when a query runs?
Question 6A. All ten nodes work until the 2% nodes finish, after which the remaining data is effectively processed by only seven nodesB. The query fails because the nodes no longer hold equal sharesC. The three nodes with 2% each take the longest to completeD. The cluster automatically redistributes the data before running the queryE. The query runs entirely on the three heaviest nodes
7. What exercise is normally run to address the Data Skew Issue after a data pipeline execution or Data Enrichment?
Question 7A. Index rebuildingB. Data standardisationC. Adding more nodes to the clusterD. Load Balancing, so data is evenly redistributed across the clusterE. Purging the oldest data
8. What is described as the first thumb rule for query performance in the era of Cloud and MPP?
Question 8A. Queries should always be limited to a single node to avoid contentionB. When a query runs, all possible compute power should be utilisedC. Data should be kept on as few nodes as possibleD. Every query should be cached before it is executedE. Compute should be added before storage
9. Which of these is NOT listed as a type of Skewed Data?
Question 9A. Skewed Right (Positive)B. Skewed Left (Negative)C. Skewed OutlierD. Skewed Narrow RangeE. Skewed Diagonal
10. A query runs exceptionally well until 99% and then hangs for hours, and adding CPU, RAM and nodes does not help. Which explanation matches this symptom?
Question 10A. The joins in the query are incorrectB. The data quality of the dataset is poorC. A Data Skew Issue: the data is unevenly distributed across the clusterD. The indexes on the tables have been disabledE. The dataset is too small to benefit from parallel processing