Not all storage is created equal: introducing Storage Buckets

The Storage Standard defines an API for persistent storage and
quota estimates, and the platform storage architecture. It’s available behind the
#enable-experimental-web-platform-features flag in Chromium 106, and we’d love your feedback.

# What problem does the storage standard aim to solve?

Traditionally, as the user runs out of storage space on their device, the data stored with APIs like
IndexedDB or localStorage gets lost without the user being able to intervene. A way to make
storage persistent is through invoking the
persist() method of the
StorageManager interface. It simultaneously requests the end user for permission and changes the
storage to be persistent once granted:

const persisted = await navigator.storage.persist();
if (persisted) {
/* Storage will not be cleared except by explicit user action. */
}

This method of asking for storage to be persisted is all or nothing. There’s no way to express more
fine-grained persistence needs. It’s all one storage bucket.

# The Storage Buckets proposal

The core idea of the Storage Buckets proposal is
granting sites the ability to create multiple storage buckets, where the browser may choose to
delete each bucket independently of other buckets. This allows developers to specify eviction
prioritization to make sure the most valuable data doesn’t get deleted.

# Use case example

To illustrate where storage buckets would come in handy, imagine an email application. It would be
unforgivable if the app lost the user’s unsent drafts that only exist on the client. In contrast, if
they are stored on a server, the user would probably be fine with some of their oldest inbox emails
to be removed from the client if their browser is under heavy storage pressure.

# Using the Storage Buckets API

# Creating a new storage bucket

A new storage bucket can be created with the open() method on the StorageBucketManager
interface.

// Create a storage bucket for emails that are synchronized with the
// server.
const inboxBucket = await navigator.storageBuckets.open('inbox');

# Creating a persisted new storage bucket

To ensure the storage bucket is persisted, you can pass durability and persisted option
arguments to the open() method:

• persisted determines if the storage bucket should be persisted or not. The allowed values are
  either false (default) or true.

• durability provides a hint to the browser that helps it trade off write performance against a
  reduced risk of data loss in the event of power failures. The allowed values are 'relaxed'
  (default) or 'strict':

  • 'strict' buckets attempt to minimize the risk of data loss on power failure. This may come at
    the cost of reduced performance, meaning that writes may take longer to complete, might impact
    overall system performance, may consume more battery power, and may wear out the storage device
    faster.
  • 'relaxed' buckets may «forget» writes that were completed in the last few seconds, when a
    power loss occurs. In return, writing data to these buckets may have better performance
    characteristics, and may allow a battery charge to last longer, and may result in longer storage
    device lifetime. Also, power failures will not lead to data corruption at a higher rate than for
    'strict' buckets.

// Create a storage bucket for email drafts that only exist on the client.
const draftsBucket = await navigator.storageBuckets.open('drafts', {
durability: 'strict', // Or `'relaxed'`.
persisted: true, // Or `false`.
});

# Accessing the storage APIs from a storage bucket

Each storage bucket is associated with storage APIs, for example,
IndexedDB, the
Cache interface, or the
File interface. These storage APIs work as per
the usual, just that the entry point is from the StorageBucket interface, for example,
StorageBucket.indexedDB.

const inboxDb = await new Promise(resolve => {
const request = inboxBucket.indexedDB.open('messages');
  request.onupgradeneeded = () => { /* migration code */ };
  request.onsuccess = () => resolve(request.result);
  request.onerror = () => reject(request.error);
});

# Feedback

The Chrome team wants to hear what you think of storage buckets. Your feedback on the
Storage Buckets proposal is warmly welcomed.
Please provide feedback by commenting on existing or filing new
GitHub Issues.

# Useful resources

• Storage Buckets explainer
• Storage Buckets repository
• ChromeStatus entry
• Chromium bug
• Blink component
  Blink > Storage > Buckets
• Mozilla Standards Position
• TAG review