Memgraph - Frequently asked questions
Memgraph 101
What is Memgraph?
Memgraph is an open-source in-memory graph database built for teams that expect highly performant, advanced analytical insights - as compatible with your current infrastructure as Neo4j (but up to 120x faster). Memgraph is powered by a query engine built in C/C++ to handle real-time use cases at an enterprise scale. Memgraph supports strongly-consistent ACID transactions and uses the standardized Cypher query language over Bolt protocol for structuring, manipulating, and exploring data.
What are the benefits of being an in-memory graph database?
When data is stored on disk, the computer has to physically read it from the disk and transfer it to the RAM before it can be processed. This process is relatively slow because it involves several physical processes, such as seeking the right location on the disk and waiting for the data to be read. Writing the data is also much slower for the same reasons.
Storing data in the computer's RAM eliminates the need for these physical processes, and data can be accessed and added almost instantly.
Therefore, in-memory graph databases are ideal for applications requiring fast data processing, real-time analytics, and quick response times.
What use cases is Memgraph best suited for?
Memgraph is best suited for use cases with complex data relationships that require real-time processing and high scalability.
How do I know my use case has complex data relationships?
In relational databases, complex data relationships arise when data from different tables is related or somehow interconnected. Because data is spread across multiple tables, querying it requires hopping from one table to other and joining it with slow and resource-intensive join operations.
The complexity of join operations can increase exponentially as the number of tables increases and as the links between various tables are no longer neatly structured following a clearly set pattern. It is no longer sufficient to join just two or three tables but hop through more than seven tables to find the correct link between the data and gain valuable analytics.
Examples of complex data are deep hierarchical relationships between data, such as parent-child relationships or many-to-many relationships between different tables.
How does Memgraph compare to other graph databases regarding performance?
Memgraph is designed to be a high-performance graph database, and it typically outperforms many other graph databases in terms of speed and scalability. Key factors contributing to Memgraph's performance are its in-memory architecture and a performant query engine written in C++. Memgraph also offers a variety of tools and features to help optimize query performance, including label and label-property indexes and a custom visualization library. Check our benchmark comparing Memgraph and Neo4j.
Is Memgraph a distributed database?
At the moment, Memgraph does not support running and storing data across multiple physical locations, but the next version of Memgraph, Memgraph 3.0 will enable horizontal scaling. Check out plans for Memgraph 3.0 on GitHub.
How does Memgraph ensures persistency and durability?
Although Memgraph is an in-memory database, the data is persistent and durable. This means data will not be lost upon reboot.
Memgraph uses two mechanisms to ensure the durability of stored data and make disaster recovery possible:
- write-ahead logging (WAL)
- periodic snapshot creation
Each database modification is recorded in a log file before being written to the database. Therefore, the log file contains all steps needed to reconstruct the DB’s most recent state. Memgraph also periodically takes snapshots during runtime to write the entire data storage to the drive. On startup, the database state is recovered from the most recent snapshot file. The timestamp of the snapshot is compared with the latest update recorded in the WAL file and, if the snapshot is less recent, the state of the DB will be completely recovered using the WAL file.
If you are using Memgraph with Docker, be sure to specify a volume for data persistence.
How does Memgraph ensure high availability?
Memgraph ensures high availability by using replication. Replication involves replicating data from one MAIN instance to one or several REPLICA instances. If a MAIN instance fails, another REPLICA instance can be upgraded and serve as the MAIN instance, thus ensuring continuous data availability.
Does Memgraph support multitenancy?
Memgraph doesn't support multitenancy, but it is planned that the next version of Memgraph, Memgraph 3.0 does. Check out plans for Memgraph 3.0 on GitHub.
How many cores does Memgraph utilizes?
Memgraph is designed to utilize all available CPU cores on a machine to process queries and perform other operations in parallel, significantly improving performance and reducing query response times.
What are the minimum and recommended requirements to run a on-premise instance?
To run Memgraph on-premise, you require an Intel Xeon, AMD Opteron/Epyc, ARM machines, Apple M1, Amazon Graviton server or desktop processor, at least 1 GB of RAM and disk and at least 1 vCPU. We recommend using a server processor, at least 16 GB of ECC RAM, the same amount of disk storage and at least 8 vCPUs or 4 physical cores.
How much RAM do I need for my graph?
We recommend twice as many GB of RAM as the data size. If you have 8 GB of data, we recommend having at least 16 GB of RAM. Of course, the actual memory needs depend on the complexity of executed queries. The more graph objects query needs to return as a result, the more RAM will be required. To calculate the Memgraph RAM instance requirements based on your data, check out how Memgraph uses memory.
Are there any graph size limits?
Memgraph vertically scales effortlessly up to 1B nodes and 10B edges. The only limit is the size of your RAM. We recommend twice as many GB of RAM as the data size. If you have 8 GB of data, we recommend having at least 16 GB of RAM. Of course, the actual memory needs depend on the complexity of executed queries. The more graph objects query needs to return as a result, the more RAM will be required.
What is the difference between Memgraph and Memgraph Platform?
There are three official Docker images for Memgraph:
memgraph/memgraph- the most basic MemgraphDB instance.memgraph/memgraph-mage- the image contains a MemgraphDB instance together with all the newest MAGE modules and graph algorithms.memgraph/memgraph-platform- the image contains MemgraphDB, Memgraph Lab, mgconsole and MAGE. Once started, mgconsole will be opened in the terminal, while Memgraph Lab is available athttp://localhost:3000.
The MAGE graph algorithm library includes NVIDIA
cuGraph GPU-powered graph algorithms. To
use them, you need a specific kind of memgraph-mage image, so check the
documentation or
DockerHub
for tags.
Do I need to define a schema before importing data?
It is not necessary to define any data schema to import data. Data will be imported into the database regardless of the number of properties and their types. You can enforce property uniqueness and existence.
Are there any educational materials available?
You can try running queries on preloaded datasets in Memgraph Playground. If you need help with Cypher queries, check out the Cypher manual. We've prepared tutorials and how-to guides to help you navigate Memgraph more easily. We also offer data modeling and Cypher e-mail courses or watch one of our webinars. You can even deep dive into code with Memgraph's CTO -> Code with Buda. For all the other questions and help, fell free to join our community.
Can I try out Memgraph Enterprise before making a decision?
Yes, Memgraph offers a free q30-day Memgraph Enterprise Trial. Send a request via the form.
Does Memgraph offer professional services such as data modelling, development, integration and similar?
It depends on the scope of the project and the requirements. Contact us for more information.
MemgraphDB
What is the fastest way to import data into Memgraph?
Currently, the fastest way to import data is from a CSV file with a LOAD CSV clause. LOAD CSV clause imports between 100K and 350K nodes per second and between 60K and 80K edges per second.
Other import methods include importing data from JSON and CYPHERL files, migrating from SQL and Neo4j with mgmigrate tool, or connecting to a data stream.
How to import data from MySQL or PostgreSQL?
You can migrate from MySQL or PostgreSQL using the mgmigrate tool.
What file formats does Memgraph support for import?
You can import data from CSV, JSON or CYPHERL files.
CSV files can be imported in on-premise instances using the LOAD CSV clause.
Local JSON files and files on a remote address can be imported in on-premise instances using a json_util module from the MAGE library. On a Cloud instance, data from JSON files can be imported only from a remote address.
CYPHERL file contains Cypher queries necessary for creating nodes and relationships.
What data formats does Memgraph support for export?
You can export data to JSON or CYPHERL files. Query results can be exported to a CSV file.
Data can be exported to a JSON file from on-premise instances using a export_util module from the MAGE library. The same module can be used to export query results to a CSV file.
CYPHERL file contains Cypher queries necessary for creating nodes and relationships and you can export files via Memgraph Lab.
Can Memgraph database ingest streaming data?
Yes, you can connect your instance to Kafka, Redpanda or Pulsar streams and ingest data. You will need to write a transformation module that will instruct Memgraph on how to transform the incoming messages and consume them correctly.
Is data automatically indexed during import?
No, data is not automatically indexed during import. You need to create a label or label-property indexes manually once the import is finished.
What languages can be used to communicate with the database?
At the moment, you can connect to a Memgraph instance using the Bolt protocol and query the database using C#, C/C++, Go, Haskell, Java, JavaScript, Node.js, PHP, Python, Ruby, and Rust.
Can I create logically separated graphs within the same database instance?
You can create logically separated graphs within the same instance by using different labels. Each node can have multiple labels and the cost of labels is 8B per label (but the memory is allocated dynamically, so 3 labels take up as much memory as 4, and 5-7 labels take as much space as 8, etc.) You can use the same technique to save multilayer networks.
Can I run MAGE modules and algorithms on just a part of the graph/subgraph?
You can run MAGE modules and algorithms on subgraphs by using the project() function.
How can I visualize query results?
You can use Memgraph Lab, a visual user interface that enables you to:
- visualize graph data using the Orb library
- write and execute Cypher queries
- import and export data
- view and optimize query performance
- develop query modules in Python
- manage connections to streams.
Does replication affect performance?
Replication should not in any way affect the performance of your database instance.
How can I check storage information?
You can check storage information by running the SHOW STORAGE INFO; that will provide information about the number of stored nodes and relationships and memory usage.
Where does Memgraph save or preview logs?
By default, Memgraph saves the log at /var/log/memgraph/memgraph.log.
Accessing logs depends on how you've started Memgraph, so check the how-to
guide about accessing logs.
You can check the logs using Memgraph Lab (the visual interface). Memgraph Lab listens to logs on the 7444 port. You can also use this web socket port 7444 and listen to the logs from your custom system.
Log level and location can be modified using configuration flags.
Do I need to know Cypher to query the database?
You don't need to know Cypher to query the database. You can use GQLAlchemy, an Object Graph Mapper (OGM). OGM provides a developer-friendly workflow for writing object-oriented notation to communicate to a graph database. Instead of writing Cypher queries, you can write Python code, which the OGM will automatically translate into Cypher queries. It supports both Memgraph and Neo4j.
Cypher
Are there any differences in Cypher implementation between Memgraph and Neo4j?
Although we tried to implement openCypher query language as closely to the language reference as possible, we made some changes that can enhance the user experience. You can find the differences listed in the Cypher manual.
Can I expand Cypher query language with custom procedures?
Yes, you can expand the Cypher query language with custom procedures grouped in query modules. Modules can be written in C/C++ and Python (which also has a mock API). For more details, check out the documentation on query modules.