DAS-C01 · Question #83
A company wants to research user turnover by analyzing the past 3 months of user activities. With millions of users, 1.5 TB of uncompressed data is generated each day. A 30-node Amazon Redshift…
The correct answer is B. Keep the data from the last 90 days in Amazon Redshift. The cost optimization insight is separating hot and cold workloads. The 30-node Redshift cluster is sized for the frequent 3-month (90-day) queries-this is the right-sized cluster for daily performance needs. The yearly historical analysis runs only once per week, so it does…
Question
A company wants to research user turnover by analyzing the past 3 months of user activities. With millions of users, 1.5 TB of uncompressed data is generated each day. A 30-node Amazon Redshift cluster with 2.56 TB of solid state drive (SSD) storage for each node is required to meet the query performance goals. The company wants to run an additional analysis on a year's worth of historical data to examine trends indicating which features are most popular. This analysis will be done once a week. What is the MOST cost-effective solution?
Options
- AIncrease the size of the Amazon Redshift cluster to 120 nodes so it has enough storage capacity
- BKeep the data from the last 90 days in Amazon Redshift.
- CKeep the data from the last 90 days in Amazon Redshift.
- DResize the cluster node type to the dense storage node type (DS2) for an additional 16 TB
How the community answered
(42 responses)- A17% (7)
- B45% (19)
- C31% (13)
- D7% (3)
Explanation
The cost optimization insight is separating hot and cold workloads. The 30-node Redshift cluster is sized for the frequent 3-month (90-day) queries-this is the right-sized cluster for daily performance needs. The yearly historical analysis runs only once per week, so it does not justify permanently expanding cluster capacity. The most cost-effective answer keeps 90 days of data in Redshift for fast daily queries, stores the full year of historical data in Amazon S3 in a columnar format (e.g., Parquet), and uses Amazon Redshift Spectrum to query the S3 data for the weekly trend analysis. Spectrum allows Redshift to query exabytes of S3 data without loading it into the cluster, paying only per-query rather than for always-on cluster nodes. Scaling to 120 nodes (A) or switching to DS2 nodes (D) would cost significantly more for a once-weekly workload.
Topics
Community Discussion
No community discussion yet for this question.