storr.tiledb

A TileDB Embedded-based driver for storr, a key-value store with content-addressable storage for R objects.

Overview

storr.tiledb is an R package that extends the storr ecosystem with a TileDB-based driver and custom storr-like classes that leverage the advantages of the underlying backend.

Key features

• Fast key-value operations using TileDB methods
• Additional faster serialization formats through qs2 package¹
• Per key metadata: notes and Time-To-Live (TTL) expiration timestamps next to key-value pairs
• Async/parallel operations with mirai
• In-memory caching layers with hash tables
• Native cloud storage support (AWS S3, Azure Blob, Google Cloud Storage)
• Data versioning (time-traveling) and encryption support
• Flexible schema configuration for performance tuning

Installation

Development version from GitHub:

# pak
pak::pkg_install("cgiachalis/storr.tiledb")

# remotes
remotes::install_github("cgiachalis/storr.tiledb")

From r-universe:

# Install 'storr.tiledb' in R:
install.packages('storr.tiledb', repos = c('https://cgiachalis.r-universe.dev'))

Quick start

library(storr.tiledb)

# Temp URI path
uri <- tempfile()

# Set up storr
sto <- storr_tiledb(uri, init = TRUE)

# Set
sto$set("mykey1", list(a = 1))
sto$set("mykey2", "abc")

# Get
sto$get("mykey2")
 [1] "abc"
 
# List all keys
sto$list()
[1] "mykey1" "mykey2"

# Del
sto$del("mykey1")

Documentation

For more detailed information, visit the full documentation on GitHub Pages.

• Get Started - Quick reference to basic operations
• API Usage - Learn about storr.tiledb operations through examples
• Data Model - Overview of TileDB driver data model

Alternative storr drivers

• storr_enviroment() - In-memory with R environments
• storr_rds() - RDS file format
• storr_dbi() - DBI interface
• storr_redis_api() - Redis through redux
• storr_thor() - LMDB Lightning Memory-Mapped Database via thor

Footnotes

1. qs2 package is soft dependency, so it has to be installed to use the 'qs2' or 'qdata' serialization formats. ↩