Skip to main content

Performance and Technical Limits

Performance Features​

  • Fault-Tolerant & High Availability: The Conscia DX Engine is designed to be resilient and consistently available, ensuring reliability in various situations. The Query Nodes topology ensures multiple healthy instances for each 'region'. Additionaly, in case of any instances going into failure state, traffic to the failed nodes is stopped and redirected to healthy nodes by the DNS load balancer. At the same time, the scheduler spins up replacement query nodes in the same region to restore capacity. Healthchecks are constantly run against all query nodes. These inform the DNS loadbalancer whether to continue to send traffic and the scheduler whether a replacement query node is required to be spun up.

  • Global Edge Network Deployment: With deployment on a globally distributed edge network, the DX Engine ensures that clients interact with the server that is geographically closest to them. This approach minimizes latency, providing a swift and efficient response time.

Technical Characteristics​

  • Query Processing:

    • The DX Engine initiatively constructs a dependency tree for each incoming query. This tree outlines all the components necessary to formulate a response accurately.
    • With this dependency tree, the Engine can intelligently decide which queries to run in parallel and which to execute sequentially, optimizing the process for efficiency and speed.

  • Processing Time:

    • The engine is optimized for performance, with benchmarked processing times of less than 20ms for even the most intricate queries. It ensures that responses are not only accurate but also promptly delivered.

  • Concurrent Query Handling:

    • The DX Engine can adeptly handle hundreds of queries per second. However, to maintain optimal performance and system stability, we have set hard rate limits on the number of queries per second (QPS) per customer application:
      • Enterprise Customers: Limited to 150 QPS.
      • All Other Customers: Limited to 50 QPS.
    • Queries exceeding these set limits will receive an error message in response. These limits are instituted to ensure fair and reliable service availability for all users.

Managing Performance​

  • All customers are set up with access to the Admin Node and Query Nodes. The Admin Node is used for all configuration, management, backup/restores and queries. Production loads should always be run against the Query Nodes (which only support query endpoints, ie endpoints ending in _query). It is ok to run non-Production loads against the Admin Node. Query Nodes are deployed geographically dispersed to provide low latency query performance and high-availability. The Admin Nodes will distribute Engine Configurations to all Query Nodes with maximum latency of 60 seconds.

For further details or clarification on the Conscia DX Engine’s performance and technical limits, please don’t hesitate to reach out to our support team.