We designed Algolia's architecture to ensure a 99.99% SLA (service-level agreement). All data entries are automatically indexed on three different high-end servers.
See the current status of our infrastructure at status.algolia.com, which is automatically updated each minute by external probes running on Google AppEngine.
Users perceive response time as the time it takes for their query to be processed by your search servers. Keep this response time low by choosing the datacenter closest to your users.
We also offer the option to automatically synchronize your indexes on several datacenters, so you can route users' queries to the datacenter nearest them. If interested, contact us at email@example.com.
You'll get the lowest latency by sending queries directly from your users' browsers or mobile apps to our servers.
To do this securely, Algolia lets you generate API keys that restrict access to a specific index, and set limits such as the maximum query rate for an IP address, or the key expiration time.
Algolia's server-side implementation is fully written in C++ and embedded as a module inside the Nginx high-performance HTTP server.
We built our instant search engine from the ground-up with performance in mind, with response times up to 200 times faster than Elasticsearch, and up to 20,000 times faster than SQLite FTS4. We can also run offline on mobile devices.
Check out our benchmark with Elasticsearch.