Cache hit
The Query API finds a popular query in memory and returns the cached result.
Flow
Client → Web Server → Query API → Memory Cache.
Interactive diagram · architecture
Design a key-value cache to save the results of the most recent web server queries — interactive diagram
Cache miss
The API finds matching documents, gets titles and snippets, then saves the contents in the cache.
Flow
Query API → Reverse Index Service → Document Service → Memory Cache.
Cache refresh
Cached entries expire so page changes, removals, additions, and rank changes can be refreshed.
Client
What it is
The client sends a search request.
Its job here
It resolves the site name with DNS. It then sends the request to the Load Balancer.
In the primer
DNS
What it is
DNS maps a name to an address.
Its job here
It receives the client lookup. The client then sends its request to the Load Balancer.
In the primer
Load Balancer
What it is
A load balancer spreads incoming requests.
Its job here
It receives requests from the Client. It sends each request to a Web Server.
Trade-off
Load balancers add another component to operate.
In the primer
Web Server
What it is
The web server runs as a reverse proxy.
Its job here
It receives a request from the Load Balancer. It forwards the request to the Query API.
In the primer
Query API
What it is
The Query API parses and processes a search query.
Its job here
It receives the web request and checks the Memory Cache. On a miss, it asks the Reverse Index Service for matching documents.
In the primer
Reverse Index Service
What it is
This service finds documents that match a query.
Its job here
It receives the normalized query from the Query API. It ranks matching results and passes the top ones to the Document Service.
In the primer
Document Service
What it is
This service returns document titles and snippets.
Its job here
It receives top matches from the Reverse Index Service. It returns contents for the Query API to place in the Memory Cache.
In the primer
Memory Cache
What it is
The Memory Cache holds query results in memory.
Its job here
It receives lookups from the Query API. On a hit it moves the entry to the front of the LRU list and returns cached contents; on a miss, the Query API uses the Reverse Index Service.
Examples
- Redis
- Memcached
Trade-off
Limited memory means older entries need to expire.
In the primer
←/→ step · Space play/pause · R replay · Home/End jump
Detail · follows playback, or click any box or step to pin it
Close · follow playbackClick any box for an explanation, or use Next or Play to walk through the three steps.
1 · Cache hit
Popular queries return from the Memory Cache.
The API updates the entry’s position at the front of the LRU list and returns cached contents.
2 · Cache miss
The API finds, ranks, and enriches matching documents.
The result is placed at the front of the Memory Cache’s LRU list.
3 · Cache refresh
A time to live refreshes cached entries.
Use it when page contents, pages, or page ranks change.