Soru

Zorluk: Çok zorHigh-Performing Data Ingestion and Transformation Solutions

A gaming company is developing a real-time multiplayer gaming analytics platform that receives gameplay telemetry from 800,000800,000 active concurrent players. The telemetry data stream is sent in JSON format, averaging 15 MB/s15\text{ MB/s} with peak volumes reaching 60 MB/s60\text{ MB/s} during global events. The platform must ingest this telemetry with sub-second latency, perform stateful anomaly detection and metrics aggregation over a sliding window of 5 minutes5\text{ minutes}, and save the transformed output in Apache Parquet format to Amazon S3 for long-term historical query access. The architecture must scale dynamically to handle peak volumes without manual intervention or administrative overhead.

Which combination of actions will meet these requirements? (Select two.)

  1. Configure an Amazon Kinesis Data Stream with on-demand capacity mode to ingest the incoming gaming telemetry.Cevap
  2. Deploy an Amazon Managed Service for Apache Flink application to process the telemetry stream, calculate sliding-window metrics, and output to an Amazon Data Firehose delivery stream for Parquet conversion and delivery to Amazon S3.Cevap
  3. C
    Configure an Amazon Kinesis Data Stream with provisioned capacity mode set to 1010 shards to ingest the telemetry, and use Kinesis Client Library applications on Amazon EC2 to calculate the metrics.
  4. D
    Configure an Amazon SQS standard queue to ingest the telemetry, and run containerized consumer applications on AWS Fargate to compute the sliding-window metrics.
  5. E
    Configure an Amazon Data Firehose delivery stream with a buffer interval of 300 seconds300\text{ seconds} to ingest the telemetry, perform the sliding-window aggregation using an inline AWS Lambda function, and deliver the data directly to Amazon S3.

Cevap

To meet the requirements, the solutions architect should configure an Amazon Kinesis Data Stream in on-demand capacity mode for the ingestion layer, and deploy an Amazon Managed Service for Apache Flink application for stateful sliding-window processing, sending the output to Amazon Data Firehose for format conversion and delivery to Amazon S3.
The correct solution involves configuring Amazon Kinesis Data Streams in on-demand mode and using Amazon Managed Service for Apache Flink with Amazon Data Firehose. Kinesis Data Streams in on-demand mode scales dynamically to handle write throughput up to 200 MB/s200\text{ MB/s} and 200,000200,000 records per second, which comfortably supports the peak throughput of 60 MB/s60\text{ MB/s} without manual administrative overhead. Amazon Managed Service for Apache Flink provides a fully managed, serverless platform for stateful stream processing, making it ideal for performing computations like 55-minute sliding-window aggregations. By routing the output of Flink to Amazon Data Firehose, the platform can utilize Firehose's native integration with AWS Glue Schema Registry to convert the JSON data to Apache Parquet format before writing it to Amazon S3, satisfying all performance, latency, and operational requirements.

Adım Adım Çözüm

1
Determine the ingestion layer capability and scaling needs.
Peak traffic is 60 MB/s60\text{ MB/s} and average is 15 MB/s15\text{ MB/s}. A Kinesis Data Stream in on-demand mode can handle up to 200 MB/s200\text{ MB/s} of write throughput automatically, which fulfills the dynamic scaling and zero administrative overhead requirements.
Provisioned mode with a fixed shard count would either throttle or require complex custom scaling scripts.
2
Identify the stateful transformation requirements.
A 55-minute sliding window requires maintaining state across events in overlapping time windows.
Amazon Managed Service for Apache Flink is the native serverless solution for stateful, low-latency stream analytics.
3
Design the output conversion and storage strategy.
Output must be stored as Apache Parquet in Amazon S3.
Connecting the Flink output to Amazon Data Firehose allows serverless transformation from JSON to Parquet (integrating with AWS Glue Schema Registry) before writing to Amazon S3.

Anahtar Kavram

Serverless high-throughput real-time stream ingestion and stateful time-window transformation using Kinesis Data Streams, Managed Service for Apache Flink, and Amazon Data Firehose.
Bu soruyu puanla