Monitoring the Internet Computer
Abstract
The Internet Computer (IC) is a distributed platform for Web3 applications, spanning over 1,200 nodes worldwide. We present results on applying runtime monitoring to the IC. We use the MonPoly monitor and its expressive policy language with quantifiers over infinite domains, aggregations, and past and future operators. We formalize complex policies that cover common kinds of production incidents and IC-specific protocol properties, including malicious behaviors and infrastructure outages. Using these policies, we evaluate MonPoly’s performance in a large-scale case study that includes logs from both production and testing environments. We find, for example, that MonPoly performs well on testing logs, and that half of our policies applicable to production logs can be monitored in an online setting. Overall, our policies and IC traces constitute a new benchmark for first-order temporal logic monitors.