DBMS > Microsoft Azure Cosmos DB vs. Milvus vs. Spark SQL vs. TerarkDB

System Properties Comparison Microsoft Azure Cosmos DB vs. Milvus vs. Spark SQL vs. TerarkDB

Please select another system to include it in the comparison.

Editorial information provided by DB-Engines

Name

Description

Globally distributed, horizontally scalable, multi-model database service

A DBMS designed for efficient storage of vector data and vector similarity searches

Spark SQL is a component on top of 'Spark Core' for structured data processing

A key-value store forked from RocksDB with advanced compression algorithms. It can be used standalone or as a storage engine for MySQL and MongoDB

Primary database model

Document store
Graph DBMS
Key-value store
Wide column store

Vector DBMS

Relational DBMS

Key-value store

Secondary database models

Spatial DBMS

DB-Engines Ranking measures the popularity of database management systems
Trend Chart

Score	29.04
Rank	#27	Overall
	#4	Document stores
	#2	Graph DBMS
	#3	Key-value stores
	#3	Wide column stores

Score	2.31
Rank	#113	Overall
	#3	Vector DBMS

Score	18.96
Rank	#33	Overall
	#20	Relational DBMS

Score	0.00
Rank	#383	Overall
	#60	Key-value stores

Website

azure.microsoft.com/services/cosmos-db

milvus.io

spark.apache.org/sql

github.com/bytedance/terarkdb

Technical documentation

learn.microsoft.com/azure/cosmos-db

milvus.io/docs/overview.md

spark.apache.org/docs/latest/sql-programming-guide.html

bytedance.larkoffice.com/docs/doccnZmYFqHBm06BbvYgjsHHcKc

Developer

Microsoft

Apache Software Foundation

ByteDance, originally Terark

Initial release

2014

2019

2014

2016

Current release

2.3.4, January 2024

3.5.0 ( 2.13), September 2023

License

Commercial or Open Source

commercial

Open Source

Apache Version 2.0

Open Source

Apache 2.0

commercial

restricted open source version available

Cloud-based only

Only available as a cloud service

yes

DBaaS offerings (sponsored links)

Database as a Service

Providers of DBaaS offerings, please contact us to be listed.

Zilliz Cloud – Cloud-native service for Milvus

Implementation language

C++, Go

Scala

C++

Server operating systems

hosted

Linux
macOS

10.14 or later
Windows

with WSL 2 enabled

Linux
OS X
Windows

Data scheme

schema-free

yes

schema-free

Typing

predefined data types such as float or date

yes

JSON types

Vector, Numeric and String

yes

XML support

Some form of processing data in XML format, e.g. support for XML data structures, and/or support for XPath, XQuery or XSLT.

Secondary indexes

yes

All properties auto-indexed by default

SQL

Support of SQL

SQL-like query language

SQL-like DML and DDL statements

APIs and other access methods

DocumentDB API
Graph API (Gremlin)
MongoDB API
RESTful HTTP API
Table API

RESTful HTTP API

JDBC
ODBC

C++ API
Java API

Supported programming languages

.Net
C#
Java
JavaScript
JavaScript (Node.js)
MongoDB client drivers written for various programming languages
Python

C++
Go
Java
JavaScript (Node.js)
Python

Java
Python
R
Scala

C++
Java

Server-side scripts

Stored procedures

JavaScript

Triggers

JavaScript

Partitioning methods

Methods for storing different data on different nodes

Sharding

Implicit feature of the cloud service

Sharding

yes, utilizing Spark Core

none

Replication methods

Methods for redundantly storing data on multiple nodes

yes

Implicit feature of the cloud service

none

MapReduce

Offers an API for user-defined Map/Reduce methods

with Hadoop integration

Integration with Hadoop/HDInsight on Azure*

Consistency concepts

Methods to ensure consistency in a distributed system

Bounded Staleness
Consistent Prefix
Eventual Consistency
Immediate Consistency

Consistency level configurable on request level
Session Consistency

Bounded Staleness
Eventual Consistency
Immediate Consistency
Session Consistency
Tunable Consistency

Foreign keys

Referential integrity

Transaction concepts

Support to ensure data integrity after non-atomic manipulations of data

Multi-item ACID transactions with snapshot isolation within a partition

Concurrency

Support for concurrent manipulation of data

yes

Durability

Support for making data persistent

yes

In-memory capabilities

Is there an option to define some or all structures to be held in-memory only.

yes

User concepts

Access control

Access rights can be defined down to the item level

Role based access control and fine grained access rights

More information provided by the system vendor

Specific characteristics

Milvus is an open-source and cloud-native vector database built for production-ready...
» more

Competitive advantages

Highly available, versatile, and robust with millisecond latency. Supports batch...
» more

Typical application scenarios

RAG: retrieval augmented generation Video media : video understanding, video deduplication....
» more

Key customers

Milvus is trusted by thousands of enterprises, including PayPal, eBay, IKEA, LINE,...
» more

Market metrics

As of January 2024, 25k+ GitHub stars 10M+ downloads and installations 3k+ enterprise...
» more

Licensing and pricing models

Milvus was released under the open-source Apache License 2.0 in October 2019. Fully-managed...
» more

We invite representatives of system vendors to contact us for updating and extending the system information,
and for displaying vendor-provided information such as key customers, competitive advantages and market metrics.