Description
gokv contains a simple interface as key-value store abstraction as well as many implementations for Go:
- Local in-memory: Go sync.Map, Go map , FreeCache, BigCache
- Embedded: bbolt, BadgerDB, LevelDB / goleveldb, Local files
- Distributed store: Redis, Consul, etcd, Apache ZooKeeper
- Distributed cache: Memcached
- Cloud: Amazon DynamoDB, Amazon S3 (and S3-compatible services like Google Cloud Storage and self-hosted solutions like Minio), Azure Table Storage, Google Cloud Datastore, Alibaba Cloud Table Store
- SQL: MySQL, PostgreSQL
- NoSQL: MongoDB
- "NewSQL": CockroachDB
- Multi-model: Apache Ignite
gokv alternatives and similar packages
Based on the "Database" category.
Alternatively, view gokv alternatives based on common mentions on social networks and blogs.
-
cockroach
CockroachDB — the cloud native, distributed SQL database designed for high availability, effortless scale, and control over data placement. -
tidb
TiDB - the open-source, cloud-native, distributed SQL database designed for modern applications. -
groupcache
groupcache is a caching and cache-filling library, intended as a replacement for memcached in many cases. -
TinyGo
Go compiler for small places. Microcontrollers, WebAssembly (WASM/WASI), and command-line tools. Based on LLVM. -
bytebase
The GitHub/GitLab for database DevSecOps. World's most advanced database DevSecOps solution for Developer, Security, DBA and Platform Engineering teams. -
immudb
immudb - immutable database based on zero trust, SQL/Key-Value/Document model, tamperproof, data change history -
go-cache
An in-memory key:value store/cache (similar to Memcached) library for Go, suitable for single-machine applications. -
buntdb
BuntDB is an embeddable, in-memory key/value database for Go with custom indexing and geospatial support -
pREST
PostgreSQL ➕ REST, low-code, simplify and accelerate development, ⚡ instant, realtime, high-performance on any Postgres application, existing or new -
xo
Command line tool to generate idiomatic Go code for SQL databases supporting PostgreSQL, MySQL, SQLite, Oracle, and Microsoft SQL Server -
nutsdb
A simple, fast, embeddable, persistent key/value store written in pure Go. It supports fully serializable transactions and many data structures such as list, set, sorted set. -
lotusdb
Most advanced key-value database written in Go, extremely fast, compatible with LSM tree and B+ tree. -
gocraft/dbr (database records)
Additions to Go's database/sql for super fast performance and convenience.
CodeRabbit: AI Code Reviews for Developers
Do you think we are missing an alternative of gokv or a related project?
Popular Comparisons
README
gokv
Simple key-value store abstraction and implementations for Go
Contents
Features
Simple interface
Note: The interface is not final yet! See Project status for details.
type Store interface {
Set(k string, v interface{}) error
Get(k string, v interface{}) (found bool, err error)
Delete(k string) error
Close() error
}
There are detailed descriptions of the methods in the docs and in the code. You should read them if you plan to write your own gokv.Store
implementation or if you create a Go package with a method that takes a gokv.Store
as parameter, so you know exactly what happens in the background.
Implementations
Some of the following databases aren't specifically engineered for storing key-value pairs, but if someone's running them already for other purposes and doesn't want to set up one of the proper key-value stores due to administrative overhead etc., they can of course be used as well. In those cases let's focus on a few of the most popular though. This mostly goes for the SQL, NoSQL and NewSQL categories.
Feel free to suggest more stores by creating an issue or even add an actual implementation - .
For differences between the implementations, see [Choosing an implementation](docs/choosing-implementation.md).
For the Godoc of specific implementations, see https://pkg.go.dev/github.com/philippgille/gokv#section-directories.
- Local in-memory
- Embedded
- [X] bbolt (formerly known as Bolt / Bolt DB)
- [X] BadgerDB
- [X] LevelDB / goleveldb
- [X] Local files (one file per key-value pair, with the key being the filename and the value being the file content)
- Distributed store
- [X] Redis
- [X] Consul
- [X] etcd
- [X] Apache ZooKeeper
- [ ] TiKV
- Distributed cache (no presistence by default)
- Cloud
- [X] Amazon DynamoDB
- [X] Amazon S3 / Google Cloud Storage / Alibaba Cloud Object Storage Service (OSS) / DigitalOcean Spaces / Scaleway Object Storage / OpenStack Swift / Ceph / Minio / ...
- [ ] Azure Cosmos DB
- [X] Azure Table Storage
- [X] Google Cloud Datastore
- [ ] Google Cloud Firestore
- [X] Alibaba Cloud Table Store
- SQL
- [X] MySQL
- [X] PostgreSQL
- NoSQL
- [X] MongoDB
- [ ] Apache Cassandra
- "NewSQL"
- [X] CockroachDB
- [ ] TiDB
- Multi-model
- [X] Apache Ignite
- [ ] ArangoDB
- [ ] OrientDB
Again:
For differences between the implementations, see [Choosing an implementation](docs/choosing-implementation.md).
For the Godoc of specific implementations, see https://pkg.go.dev/github.com/philippgille/gokv#section-directories.
Value types
Most Go packages for key-value stores just accept a []byte
as value, which requires developers for example to marshal (and later unmarshal) their structs. gokv
is meant to be simple and make developers' lifes easier, so it accepts any type (with using interface{}
as parameter), including structs, and automatically (un-)marshals the value.
The kind of (un-)marshalling is left to the implementation. All implementations in this repository currently support JSON and gob by using the encoding
subpackage in this repository, which wraps the core functionality of the standard library's encoding/json
and encoding/gob
packages. See Marshal formats for details.
For unexported struct fields to be (un-)marshalled to/from JSON/gob, the respective custom (un-)marshalling methods need to be implemented as methods of the struct (e.g. MarshalJSON() ([]byte, error)
for custom marshalling into JSON). See Marshaler and Unmarshaler for JSON, and GobEncoder and GobDecoder for gob.
To improve performance you can also implement the custom (un-)marshalling methods so that no reflection is used by the encoding/json
/ encoding/gob
packages. This is not a disadvantage of using a generic key-value store package, it's the same as if you would use a concrete key-value store package which only accepts []byte
, requiring you to (un-)marshal your structs.
Marshal formats
This repository contains the subpackage encoding
, which is an abstraction and wrapper for the core functionality of packages like encoding/json
and encoding/gob
. The currently supported marshal formats are:
- [X] JSON
- [X] gob
More formats will be supported in the future (e.g. XML).
The stores use this encoding
package to marshal and unmarshal the values when storing / retrieving them. The default format is JSON, but all gokv.Store
implementations in this repository also support gob as alternative, configurable via their Options
.
The marshal format is up to the implementations though, so package creators using the gokv.Store
interface as parameter of a function should not make any assumptions about this. If they require any specific format they should inform the package user about this in the GoDoc of the function taking the store interface as parameter.
Differences between the formats:
- Depending on the struct, one of the formats might be faster
- Depending on the struct, one of the formats might lead to a lower storage size
- Depending on the use case, the custom (un-)marshal methods of one of the formats might be easier to implement
Roadmap
- Benchmarks!
- CLI: A simple command line interface tool that allows you create, read, update and delete key-value pairs in all of the
gokv
storages - A
combiner
package that allows you to create agokv.Store
which forwards its call to multiple implementations at the same time. So for example you can usememcached
ands3
simultaneously to have 1) super fast access but also 2) durable redundant persistent storage. - A way to directly configure the clients via the options of the underlying used Go package (e.g. not the
redis.Options
struct ingithub.com/philippgille/gokv
, but instead theredis.Options
struct ingithub.com/go-redis/redis
)- Will be optional and discouraged, because this will lead to compile errors in code that uses
gokv
when switching the underlying used Go package, but definitely useful for some people
- Will be optional and discouraged, because this will lead to compile errors in code that uses
- More stores (see stores in Implementations list with unchecked boxes)
- Maybe rename the project from
gokv
toSimpleKV
? - Maybe move all implementation packages into a subdirectory, e.g.
github.com/philippgille/gokv/store/redis
?
Usage
First, download the module you want to work with:
- For example when you want to work with the
gokv.Store
interface:go get github.com/philippgille/gokv@latest
- For example when you want to work with the Redis implementation:
go get github.com/philippgille/gokv/redis@latest
Then you can import and use it.
Every implementation has its own Options
struct, but all implementations have a NewStore()
/ NewClient()
function that returns an object of a sctruct that implements the gokv.Store
interface. Let's take the implementation for Redis as example, which is the most popular distributed key-value store.
package main
import (
"fmt"
"github.com/philippgille/gokv"
"github.com/philippgille/gokv/redis"
)
type foo struct {
Bar string
}
func main() {
options := redis.DefaultOptions // Address: "localhost:6379", Password: "", DB: 0
// Create client
client, err := redis.NewClient(options)
if err != nil {
panic(err)
}
defer client.Close()
// Store, retrieve, print and delete a value
interactWithStore(client)
}
// interactWithStore stores, retrieves, prints and deletes a value.
// It's completely independent of the store implementation.
func interactWithStore(store gokv.Store) {
// Store value
val := foo{
Bar: "baz",
}
err := store.Set("foo123", val)
if err != nil {
panic(err)
}
// Retrieve value
retrievedVal := new(foo)
found, err := store.Get("foo123", retrievedVal)
if err != nil {
panic(err)
}
if !found {
panic("Value not found")
}
fmt.Printf("foo: %+v", *retrievedVal) // Prints `foo: {Bar:baz}`
// Delete value
err = store.Delete("foo123")
if err != nil {
panic(err)
}
}
As described in the comments, that code does the following:
- Create a client for Redis
- Some implementations' stores/clients don't require to be closed, but when working with the interface (for example as function parameter) you must call
Close()
because you don't know which implementation is passed. Even if you work with a specific implementation you should always callClose()
, so you can easily change the implementation without the risk of forgetting to add the call.
- Some implementations' stores/clients don't require to be closed, but when working with the interface (for example as function parameter) you must call
- Call
interactWithStore()
, which requires agokv.Store
as parameter. This method then:- Stores an object of type
foo
in the Redis server running onlocalhost:6379
with the keyfoo123
- Retrieves the value for the key
foo123
- The check if the value was found isn't needed in this example but is included for demonstration purposes
- Prints the value. It prints
foo: {Bar:baz}
, which is exactly what was stored before. - Deletes the value
- Stores an object of type
Now let's say you don't want to use Redis but Consul instead. You just have to make three simple changes:
- Replace the import of
"github.com/philippgille/gokv/redis"
by"github.com/philippgille/gokv/consul"
- Replace
redis.DefaultOptions
byconsul.DefaultOptions
- Replace
redis.NewClient(options)
byconsul.NewClient(options)
Everything else works the same way. interactWithStore()
is completely unaffected.
Project status
Note:
gokv
's API is not stable yet and is under active development. Upcoming releases are likely to contain breaking changes as long as the version isv0.x.y
. You should use vendoring to prevent bad surprises. This project adheres to Semantic Versioning and all notable changes to this project are documented in RELEASES.md.
Planned interface methods until v1.0.0
:
List(interface{}) error
/GetAll(interface{}) error
or similar
The interface might even change until v1.0.0
. For example one consideration is to change Get(string, interface{}) (bool, error)
to Get(string, interface{}) error
(no boolean return value anymore), with the error
being something like gokv.ErrNotFound // "Key-value pair not found"
to fulfill the additional role of indicating that the key-value pair wasn't found. But at the moment we prefer the current method signature.
Also, more interfaces might be added. For example so that there's a SimpleStore
and an AdvancedStore
, with the first one containing only the basic methods and the latter one with advanced features such as key-value pair lifetimes (deletion of key-value pairs after a given time), notification of value changes via Go channels etc. But currently the focus is simplicity, see Design decisions.
Motivation
When creating a package you want the package to be usable by as many developers as possible. Let's look at a specific example: You want to create a paywall middleware for the Gin web framework. You need some database to store state. You can't use a Go map, because its data is not persisted across web service restarts. You can't use an embedded DB like bbolt, BadgerDB or SQLite, because that would restrict the web service to one instance, but nowadays every web service is designed with high horizontal scalability in mind. If you use Redis, MongoDB or PostgreSQL though, you would force the package user (the developer who creates the actual web service with Gin and your middleware) to run and administrate the server, even if she might never have used it before and doesn't know how to configure them for high performance and security.
Any decision for a specific database would limit the package's usability.
One solution would be a custom interface where you would leave the implementation to the package user. But that would require the developer to dive into the details of the Go package of the chosen key-value store. And if the developer wants to switch the store, or maybe use one for local testing and another for production, she would need to write multiple implementations.
gokv
is the solution for these problems. Package creators use the gokv.Store
interface as parameter and can call its methods within their code, leaving the decision which actual store to use to the package user. Package users pick one of the implementations, for example github.com/philippgille/gokv/redis
for Redis and pass the redis.Client
created by redis.NewClient(...)
as parameter. Package users can also develop their own implementations if they need to.
gokv
doesn't just have to be used to satisfy some gokv.Store
parameter. It can of course also be used by application / web service developers who just don't want to dive into the sometimes complicated usage of some key-value store packages.
Initially it was developed as storage
package within the project ln-paywall to provide the users of ln-paywall with multiple storage options, but at some point it made sense to turn it into a repository of its own.
Before doing so I examined existing Go packages with a similar purpose (see Related projects), but none of them fit my needs. They either had too few implementations, or they didn't automatically marshal / unmarshal passed structs, or the interface had too many methods, making the project seem too complex to maintain and extend, proven by some that were abandoned or forked (splitting the community with it).
Design decisions
gokv
is primarily an abstraction for key-value stores, not caches, so there's no need for cache eviction and timeouts.- It's still possible to have cache eviction. In some cases you can configure it on the server, or in case of Memcached it's even the default. Or you can have an implementation-specific
Option
that configures the key-value store client to set a timeout on some key-value pair when storing it in the server. But this should be implementation-specific and not be part of the interface methods, which would require every implementation to support cache eviction.
- It's still possible to have cache eviction. In some cases you can configure it on the server, or in case of Memcached it's even the default. Or you can have an implementation-specific
- The package should be usable without having to write additional code, so structs should be (un-)marshalled automatically, without having to implement
MarshalJSON()
/GobEncode()
andUnmarshalJSON()
/GobDecode()
first. It's still possible to implement these methods to customize the (un-)marshalling, for example to include unexported fields, or for higher performance (because theencoding/json
/encoding/gob
package doesn't have to use reflection). - It should be easy to create your own store implementations, as well as to review and maintain the code of this repository, so there should be as few interface methods as possible, but still enough so that functions taking the
gokv.Store
interface as parameter can do everything that's usually required when working with a key-value store. For example, a boolean return value for theDelete
method that indicates whether a value was actually deleted (because it was previously present) can be useful, but isn't a must-have, and also it would require someStore
implementations to implement the check by themselves (because the existing libraries don't support it), which would unnecessarily decrease performance for those who don't need it. Or as another example, aWatch(key string) (<-chan Notification, error)
method that sends notifications via a Go channel when the value of a given key changes is nice to have for a few use cases, but in most cases it's not required.- > Note: In the future we might add another interface, so that there's one for the basic operations and one for advanced uses.
- Similar projects name the structs that are implementations of the store interface according to the backing store, for example
boltdb.BoltDB
, but this leads to so called "stuttering" that's discouraged when writing idiomatic Go. That's whygokv
uses for examplebbolt.Store
andsyncmap.Store
. For easier differentiation between embedded DBs and DBs that have a client and a server component though, the first ones are calledStore
and the latter ones are calledClient
, for exampleredis.Client
. - All errors are implementation-specific. We could introduce a
gokv.StoreError
type and define some constants like aSetError
or something more specific like aTimeoutError
, but non-specific errors don't help the package user, and specific errors would make it very hard to create and especially maintain agokv.Store
implementation. You would need to know exactly in which cases the package (that the implementation uses) returns errors, what the errors mean (to "translate" them) and keep up with changes and additions of errors in the package. So instead, errors are just forwarded. For example, if you use thedynamodb
package, the returned errors will be errors from the"github.com/aws/aws-sdk-go
package. - Keep the terminology of used packages. This might be controversial, because an abstraction / wrapper unifies the interface of the used packages. But:
- Naming is hard. If one used package for an embedded database uses
Path
and anotherDirectory
, then how should be name the option for the database directory? MaybeFolder
, to add to the confusion? Also, some users might already have used the packages we use directly and they would wonder about the "new" variable name which has the same meaning.
Using the packages' variable names spares us the need to come up with unified, understandable variable names without alienating users who already used the packages we use directly. - Only few users are going to switch back and forth between
gokv.Store
implementations, so most user won't even notice the differences in variable names.
- Naming is hard. If one used package for an embedded database uses
- Each
gokv
implementation is a Go module. This differs from repositories that contain a single Go module with many subpackages, but has the huge advantage that if you only want to work with the Redis client for example, thego get
will only fetch the Redis dependencies and not the huge amount of dependencies that are used across the whole repository.
Related projects
- libkv
- Uses
[]byte
as value, no automatic (un-)marshalling of structs - No support for Redis, BadgerDB, Go map, MongoDB, AWS DynamoDB, Memcached, MySQL, ...
- Not actively maintained anymore (3 direct commits + 1 merged PR in the last 10+ months, as of 2018-10-13)
- Uses
- valkeyrie
- Fork of libkv
- Same disadvantage: Uses
[]byte
as value, no automatic (un-)marshalling of structs - No support for BadgerDB, Go map, MongoDB, AWS DynamoDB, Memcached, MySQL, ...
- gokvstores
- Only supports Redis and local in-memory cache
- Not actively maintained anymore (4 direct commits + 1 merged PR in the last 10+ months, as of 2018-10-13)
- 13 stars (as of 2018-10-13)
- gokv
- Requires a
json.Marshaler
/json.Unmarshaler
as parameter, so you always need to explicitly implement their methods for your structs, and also you can't use gob or other formats for (un-)marshaling. - No support for Consul, etcd, bbolt / Bolt, BadgerDB, MongoDB, AWS DynamoDB, Memcached, MySQL, ...
- Separate repo for each implementation, which has advantages and disadvantages
- No releases (makes it harder to use with package managers like dep)
- 2-7 stars (depending on the repository, as of 2018-10-13)
- Requires a
Others:
- gladkikhartem/gokv: No
Delete()
method, no Redis, embedded DBs etc., no Git tags / releases, no stars (as of 2018-11-28) - bradberger/gokv: Not maintained (no commits in the last 22 months), no Redis, Consul etc., no Git tags / releases, 1 star (as of 2018-11-28)
- This package inspired me to implement something similar to its
Codec
.
- This package inspired me to implement something similar to its
- ppacher/gokv: Not maintained (no commits in the last 22 months), no Redis, embedded DBs etc., no automatic (un-)marshalling, 1 star (as of 2018-11-28)
- Nice CLI!
- kapitan-k/gokvstore: Not actively maintained (no commits in the last 10+ months), RocksDB only, requires cgo, no automatic (un-)marshalling, no Git tags/ releases, 1 star (as of 2018-11-28)
License
gokv
is licensed under the Mozilla Public License Version 2.0.
Dependencies might be licensed under other licenses.
*Note that all licence references and agreements mentioned in the gokv README section above
are relevant to that project's source code only.