You are managing a network of 1,000 IoT sensors at SenseTrack Solutions . Each sensor generates time series data —one metric per sensor per second, with an associated timestamp . You already have 1 TB of historical data , and expect an additional 1 GB per day . You need to support two access patterns: Access Pattern 1: Retrieve a single metric for a specific sensor at a specific timestamp, with single-digit millisecond median latency . Access Pattern 2: Run complex analytical queries , including joins , on the full dataset once per day . What is the most suitable storage strategy to meet both access needs efficiently?