High-Availability store

Efficient data fetching to prevent thundering herds!

HA-store is a wrapper for your data queries, it features:

- Smart TLRU cache for 'hot' information
- Supports multiple caching levels
- Request coalescing and batching (solves the Thundering Herd problem)
- Insightful stats and events
- Lightweight, configurable, battle-tested

Learn how you can improve your app's performance, design and resilience here!

Installing

npm install ha-store

Usage

``javascript
// Create your store
import HAStore, {caches} from 'ha-store';

const itemStore = HAStore({
resolver: getItems,
delimiter: ['language'],
caches: [
caches.inMemory({
limit: 1000, // Maximum number of cached items
ttl: 60000, // Time to live: 60 seconds
}),
],
});

// Define your resolver
function getItems(ids, params, contexts) {
// Ids will be a list of all the unique requested items
// Params will be the parameters for the request, which must be declared in the delimiter config of the store
// Contexts will be the list of originating context information

// Now perform some expensive network call or database lookup...

// Then, respond with your data formatted into this formats:
// { '123': { language: 'fr', name: 'fred' } }
}

// Now to use your store
itemStore.get('123', { language: 'fr' }, { requestId: '123' })
.then(item => / The item you requested /);

// You can even ask for more than one item at a time
itemStore.getMany(['123', '456'], { language: 'en' }, { requestId: '123' })
.then(items => / All the items you requested, in Promise.allSettled fashion /);
`

$3

HA-store has prebuilt resolvers for common use-cases:

`typescript
import HAStore, {resolvers} from 'ha-store';
import { Pool } from 'pg';

// PostgreSQL connection pool
const pool = new Pool({
host: process.env.DB_HOST || 'localhost',
database: process.env.DB_NAME || 'ha_store_example',
user: process.env.DB_USER || 'postgres',
});

const store = haStore({
resolver: resolvers.postgres({
db: pool,
table: 'items',
identifier: 'id',
}),
delimiter: ['language', 'region'],
batch: {
delay: 50,
limit: 50,
}
});
`

Options

Name | Required | Default | Description
--- | --- | --- | ---
resolver | true | - | The method to wrap, and how to interpret the returned data. Uses the format
delimiter | false | [] | The list of parameters that, when passed, generate unique results. Ex: 'language', 'view', 'fields', 'country'. These will generate different combinations of cache keys.
caches | false |

[
   <instance of a store>,
]

{
  delay: 50,
  limit: 100
}

{
  limit: 5000,
  ttl: 300000
}

{
  ttl: 0,
  keyspace: '',
  host: '0.0.0.0',
  path: null,
  connection: <instance of redis connection>
}

Event | Format | Description
--- | --- | ---
localCacheHit | | When the requested item is present in the first listed local store (usually in-memory).
cacheHit | | When the requested item is present in a store.
cacheMiss | | When the requested item not cached or coalesced and must be fetched.
coalescedHit | | When a record query successfully hooks to the promise of the same record in transit.
query |