Character Trie vs Reverse Word Trie
Evaluating CPU Execution Speed versus Memory Allocation (RAM) when processing a single network request (ads.doubleclick.net). This demonstrates why Nova Shield utilizes a Subdomain Trie architecture.
Character Trie Traditional
CPU Execution Speed
0 / 19 Ops
Iterates through every single character sequentially. High CPU overhead.
Memory Allocation (RAM)
~15 MB
Stores 1-byte character nodes. Highly memory efficient.
Subdomain Trie Nova Shield
CPU Execution Speed
0 / 3 Ops
Splits by domain intervals. Skips entire words per cycle. Ultra-low latency.
Memory Allocation (RAM)
~45 MB
Stores complete String objects in Hash Maps. Marginally higher memory footprint.
The Engineering Verdict: By sacrificing a negligible ~30 MB of modern device RAM, Nova Shield reduces CPU workload by over 84% per network request. In a proxy server handling hundreds of concurrent connections, this trade-off is essential for maintaining zero-latency routing and preserving battery life.