Design a system for real-time anomaly detection in system metrics.

Question

Anonymous · Accepted Answer

requiremements

how much data does the system need to process?

what is the customer experience?

what are latency requirments?

1. 1. entry point- api gateway routing solution route metrics reported to an
    api endpoint

2. 2. Compute - A distributed data stream solution like kafka. A server based
    compute solution that can be autoscaled reads from the kafka stream to
    perform data analysis.

3. 3. Storage - I would use a write effiecent storage solution like cassandra
    that would store data directly from the stream.

4. 4. If an anomoly occurs the compute puts a message on a distrubuted pub/sub
    queue and then that data gets picked up by another serverless compute
    solution that cuts a ticket to the customer.

5. 5. the Front end static website also subscribes to this topic to publish new
    anomolies on the website.

6. 6. There a CDN to cache part of the website at the edge but not
    notification.

Design a system for real-time anomaly detection in system metrics.

Practice this question with AI

Go Premium

Community Answers

Unlock Full Access

Practice More Questions