AWS Certified Data Analytics Specialty 2021 – Hands-On - Kinesis Scaling

The video tutorial explains how to scale Amazon Kinesis by adding or splitting shards to increase stream capacity. It covers shard merging to reduce capacity and save costs. The tutorial also discusses the impact of re-sharding on data order and how to handle it. Auto scaling is addressed, noting that it's not a native feature and requires manual implementation. Limitations of Kinesis scaling, such as the inability to perform parallel re-sharding, are highlighted.

What is the primary purpose of shard splitting in Kinesis?

• To decrease the number of shards

• To merge low-traffic shards

• To reduce data latency

• To increase stream capacity

How does shard merging help in managing Kinesis streams?

• It allows parallel resharing

• It improves data processing speed

• It helps in reducing costs by combining low-traffic shards

• It increases the number of shards

What issue might arise if a consumer reads from child shards before finishing with parent shards?

• Data might be duplicated

• Data processing speed might decrease

• Data might be received out of order

• Data might be lost

What is a key guideline to ensure data order after resharing in Kinesis?

• Read from child shards first

• Increase the number of shards

• Use multiple consumer applications

• Ensure all data from parent shards is read before moving to child shards

What is a limitation of Kinesis scaling?

• It automatically adjusts shard count

• It cannot scale up or down too quickly

• It is instantaneous

• It allows parallel resharing

How can auto-scaling be implemented in Kinesis?

• Using a native Kinesis feature

• Through manual adjustments only

• By using external tools like Lambda

• By increasing the number of shards

Why is it important to plan capacity in advance for Kinesis scaling?

• Because resharing takes time and cannot be done in parallel

• Because scaling is instantaneous

• Because it automatically adjusts to traffic

• Because resharing can be done in parallel