v4 Ultipa Grid

First Release: Q1 2023
Keywords: Demi-schema, Plug-n-play Algorithm, Product Matrix
Features:

• Product matrix
• Demi-schema
• Completed functionality of UQL
• EXTA
• Plug-n-Play algorithms
• Rich collection of graph algorithms (over 50)

Supporting Products:

• UQL v4.x
• Ultipa Manager v4.x
• Ultipa Transporter v4.x
• Ultipa Cloud
• Ultipa Deployer v4.x
• Ultipa Maker v4.x
• Ultipa Migrator v4.x
• Ultipa CLI v4.x
• Ultipa SDKs v4.x
• Ultipa Restful API
• Ultipa Playground
• Ultipa BI
• Ultipa Monitor v4.x
• Ultipa Daemon v4.x
• Ultipa Name Server v4.x

Optimizations:

• Lower memory cost
• Lighting fast k-hop, shortest path, and everything else revamped
• Advanced query optimizers
• HTAP & Grid (Name Server) architecture support
• Large enterprise grade one-click deployment
• enriched graph embedding / GNN support

Release Notes:

Ultipa Grid, the 4th generation of Ultipa Graph System, keeps creating miracles in the industry last past year by constantly surpassing its older versions. 2022 is a very productive year for Ultipa, after long time of technological accumulation, Ultipa has completed its Product Matrix, which in essence covers various comprehensive tool chains to do everything with and about Ultipa Graph. This includes several mandatory steps in a workflow starting from deploying an Ultipa sever either on premise or in the public cloud, migrating data from different types of data sources in diversified data structures, invoking graph queries and graph management via different user interfaces (CLI, Manager, SDKs, etc.), as well as some extensional applications of Ultipa server.

This Engine v4.x adopts a Demi-Schema mechanism that offers ultimate flexibility in terms of how data modeling and processing is done. Users can go either schema-strict during graph modeling and graph query, which offers superior data processing speed particularly when handling convoluted and heterogeneous types of entities and relations, or schema-free that is super flexible when working with complex graph datasets.

The Completed Functionality of UQL fully covers the features of DDL, DML, DQL and DCL, realized by employing commands of graph modelling, CRUD of metadata, indexes, tasks, users and privileges, queries on path, subgraph. The processing of query results is facilitated with a rich collection of functions and clauses in combination of alias mechanism, which allows those advanced high-dimensional data structures to be operated and returned.

The Plug-n-Play Algorithm feature has been upgraded to a new level by empowering users to not only install and update standard algorithms without system reboot – this is especially useful for enterprise setup, comparable to changing wheels while the training is fast rolling forward – but also extend the analysis capabilities of Ultipa Grid through docking one’s own graph algorithm with an Ultipa engine, which is a mission impossible until the launch of EXTA.

A highlight should be spotted on one product among all that have been published in the high-productivity fiscal year, the web application, Ultipa Cloud, a graph database as a service with which an Ultipa instance can be started in just a few minutes. Choose either a low-cost configuration of server if you are an individual user who wants to learn about graph, or an enterprise ready for a brand-new business experience brought by graph technology, or elect a server configuration competent for your business profile. It is always paid as you go!

With its last update in 2023-Q1, Ultipa Grid has boasted a variety of new features and optimizations to enhance its performance and robustness, including lower memory cost, more efficient K-Hop and Shortest Path calculations, an advanced query optimizer, and the HTAP & Grid (Name Sever) architecture, making it easier and more efficient for users to manage and analyze their business in large graph datasets.

v4.4.41

Date: April 23, 2024

1. Optimized memory usage of the graph topological structure.
2. Enhanced the efficiency of template aggregation.
3. Resolved the startup consistency issue.
4. Fixed known bugs.

v4.3.94

Date: November 17, 2023

1. Added support for extracting x and y values from the point-type property and introduced the new pointInPolygon() function.
2. Introduced trigger-related clauses.
3. Added support for the blob, decimal and set property types.
4. Expanded privilege management to the property level.
5. Introduced new string functions: trim(), ltrim(), rtrim(), left(), right(), substring(), reverse(), replace(), JSON_decode(), and JSON_merge().
6. Implemented support for encrypted storage for properties of string and text types.
7. Added string concatenation with the operator +, for example, return 'a'+'b'.
8. Implemented incremental backup functionality.
9. Added a size column to the return table of show().index() clause, with size measured in bytes.
10. Introduced the TRY prefix keyword.
11. New supported syntax: n(<filter?> as nodes) as paths return nodes{*}, paths{*}
12. Introduced new type conversion functions: toSet(), toDouble(), and toDecimal().
13. Introduced new operators: KhopTemplateCount, KhopTemplateGroupCount, TemplateCount, and OptionalTemplateCount. They are visible when using the EXPLAIN prefix.
14. Fixed known bugs.

v4.3.52

Date: June 21, 2023

1. LPA and HANP algorithms now support multiple properties.
2. Enhanced rules for template queries.
3. Improved EXTA interface performance.
4. Resolved installation issue of EXTA in ARM version.
5. Addressed abnormal crash problems and fixed known bugs.

4.2.65

Date: May 11, 2023

1. Resolved memory leak in "ultipa.lic" certificate timing monitoring.
2. Fixed out-of-memory (OOM) bug in k-hop.
3. Addressed crash issue in the Eigenvector Centrality" algorithm.

v4.3.44

Date: May 6, 2023

1. Introduced the list type to properties. Examples: create().node_property(@People, "name", "string[]") create().node_property(@People, "age", "int32[]") insert().into(@People).nodes({name:["john", "alex"]}) find().nodes({name == ["john", "alex"]}) return nodes{*}
2. Add support for null property values and null-related operations.
3. Introduced IS NULL and IS NOT NULL operators.
4. Aggregation functions now ignore null values.
5. Added ListData tag, is_null tags, and Backup interface to the proto, also added sub_types in Property.
6. Introduced the point type to properties. Examples: create().node_property(@country, "test_point", point) insert().into(@country).nodes({test_point: point({latitude:13.43, longitude:56.21})})
7. Introducted new spatial functions: point() and distance(). Example: WITH point({latitude:13.43, longitude:56.21}) AS p1 return p1 WITH point({latitude:13.43, longitude:56.21}) AS p1, point({latitude:13.10, longitude:56.41}) AS p2 RETURN distance(p1, p2)/1000 AS km
8. Added support for OPTIONAL predix for find nodes, find edges,k-hop, and k-hop template queries.
9. Refactored EXTA.
10. Renamed distinct() function to dedup(), DISTINCT is now an operator keyword.
11. Refactored EXPLAIN and PROFILE prefix functionalities, added new operators: apply, distinct, project, mergepipe, mergepipeforreturn, etc.
12. Added new startup parameter -restore to support restoring data from backup file directory.
13. Server.config configuration parameters private_addr and public_addr now support domain names.

4.2.59

Date: February 23, 2023

1. Resolved crashes related to lists.
2. Resolved bugs related to null values in aggregation functions.
3. Merged HTAP log and server log into a single file.

4.2.53

Date: February 6, 2023

1. Enhanced metadata structure of the graph to reduce memory usage.
2. Improved storage and performance of modification functions.
3. Introduced EXTA functionality that allows custom algorithm plugins.
4. Added the text type to properties which doesn't have text-length limit.
5. Introduced "Server.docker_mem_usage_path" and "Server.memory_max_limit" in config for memory limit control in Cloud version.
6. Restructured list data type to support element retrieval by index (e.g., list[0], list[1:2], list[2:]).
7. Timestamp data type now allows SDK to set time zone.
8. Revised naming rules for schema, property, and alias to support special characters and Chinese.
9. Insert, delete, update, and some query clauses now supports the limit() method and OPTIONAL prefix.
10. Added new functions: dateFormat(), toString(), range(), ifnull(), reduce().
11. Renamed some functions: date_add() to dateAdd(), date_diff() to dateDiff(), day_of_week() to dayOfWeek(), array_union() to listUnion(), starts_with() to startsWith(), ends_with() to endsWith().
12. Introduced the DEBUG prefix for performance statistics with better granularity.
13. Modified the PROFILE prefix functionality to display each clause's time cost.
14. Enhanced the display content of EXPLAIN prefix.
15. WHERE clause now supports aggregate functions.
16. Modified mathematical operators and functions to handle null values.
17. Fixed other bugs.

v3 Ultipa HTAP

First Release: 2021
Keywords: HTAP Graph Data Processing, Enterprise Stability, Upgraded UQL
Features:

• High availability server cluster
• TP and AP rules settings
• HTAP (Hybrid Transactional/Analytical Processing) support
• New storage engine
• Pluggable graph algorithms
• Algorithm execution as background task
• Algorithm execution as real-time process
• Engine index (LTE/UFE)
• Disk Index (Property index, full-text index)

Supporting Products:

• UQL v3.x
• Ultipa Manager v3.x
• Ultipa Transporter v3.x
• Ultipa CLI v3.0
• SDKs (JAVA/Python/Golang/Nodejs/C++)

Optimizations:

• Greater HDPC engine performance
• Reduced memory usage
• 90 * 24 high TPS writing and computing of 24x7 guaranteed stability & data consistency

Release Notes:

Ultipa HTAP, the third generation of Ultipa Graph System, has been continuously publishing new versions for the whole 2021, bringing with it a range of new features and optimizations. After iterations with v3.1, v3.2, and v3.3, the system now is a veritable High Availability Server Cluster ensuring maximized uptime and reliability. An outstanding breakthrough of this 3rd generation engine is its balance between transaction and analysis, the so called HTAP (Hybrid Transactional/Analytical Processing) feature that enables real-time processing of both transactional and analytical workloads.

With the feature of HTAP, these v3.x releases allow for custom TP and AP Rules Settings, tailoring the system to specific needs. Functionality of graph algorithm also come with significant updates, not only can each graph algorithm be Installed and Hot Plugged, making it easier to customize based on user needs and hardware adequacy, but the execution of any graph algorithm can now be selected between Real-time Process and Background Task to adapt to different computational loads of the algorithms.

Introducing indexes is another key improvement of these v3.x releases. The traditional techniques of index tree and revert index in full-text search scenario are employed to augment the filtering efficiency against metadata properties, by storing index trees of the target property values in the server disk. But since Memory is the Next Generation of Storage, these v3.x releases adopt the Engine Index that manipulate a copy of metadata properties by invoking commands LTE/UFE (Load To Engine/Unload From Engine), which greatly reduce I/O cost hence speeds up path query.

With a bunch of original products upgraded and adapted to the current engine version, such as Ultipa Manager v3.x and Ultipa CLI v3.x, new members are also introduced into the SDK family, developer using JAVA, Python, Golang, Nodejs, and C++ now can easily connect and manipulate an Ultipa Graph System in their projects or applications.

Ultipa HTAP has been optimized for performance and stability, with improvements to HDC engine performance and reduced memory usage. It now guarantees 90 * 24 high TPS writing and computing, ensuring maximum stability and reliability. It is well-suited for a wide range of use cases in data science, analytics, and research. With its advanced features and optimizations, it delivers unprecedented performance, scalability, and reliability.

v2 Ultipa Colossus

First Release: 2020
Keywords: UQL 1.0, 30+ Algorithms, Storage Engine
Features:

• High-density & parallel computing capability
• Native graph storage
• Chain-style query language
• Over 30 graph algorithms

Capabilities:

• Command-style UQL graph queries
• Multi-graph management
• Auto-net for networked pathfinding among entities

Supporting Products:

• UQL v2.x
• Ultipa Manager v2.x
• Ultipa Transporter v2.x
• SDKs (Golang/C++)

SDKs for Golang and C++ are also in place for this exclusive graph query language to be invoked when projects or applications are to be built by developers and an Ultipa server can be connected and operated with ease.

Ultipa Colossus retains the powerfulness and versatility from its predecessor while delivering significant improvements. With its advanced features and capabilities listed above, it is well-suited for a wide range of use cases in research, analytics, and data science.

Ultipa Colossus, the second generation of the Ultipa Graph System, is introducing exciting new features and capabilities. The upgraded High-Density & Parallel Graph Computing Engine v2.0 boosts performance and efficiency by fully leveraging the underlying infrastructure computing power, while the Graph Native Storage Engine is invented to continue leveraging the strengths of index-free adjacency data structure that enables fast and seamless data management.

Engine v2.0 highlights include support for multi-graph management, allowing storage of multiple graph datasets for improved productivity. Auto-net enhances AB-path finding by enabling batch processing of pathing queries, supporting multiple initial and terminal entities. The engine supports over 30 graph algorithms for comprehensive data analysis and visualization.

Ultipa Manager v2.x and Ultipa Transporter v2.x are the upgraded versions of their previous releases in 2019. An essential and brand-new product offered this time is UQL v2.x, a powerful command-like graph query language adapting to the engine in all aspects, enabling and simplifying graph query and management. SDKs for Golang and C++ are also in place for this exclusive graph query language to be invoked when projects or applications are to be built by developers and an Ultipa server can be connected and operated with ease.

Ultipa Colossus retains the powerfulness and versatility from its predecessor while delivering significant improvements. With its advanced features and capabilities listed above, it is well-suited for a wide range of use cases in research, analytics, and data science.

v1 Ultipa Genesis

First Release: December, 2019
Keywords: High Density Graph Computing Engine
Features:

• Ultra-deep data penetration, exponentially faster and deeper than competitors.
• Highly visualized graph DBMS for augmented query/data/results XAI (eXplainability).
• Utilizes Ultipa index-free adjacency data structure for graph computing.
• Dynamic graph pruning for improved traversal efficiency.

Capabilities:

• AB-path finding and k-hop counting in microseconds
• Automatic spreading
• 2D/3D visualization
• Loading Twitter-2010 within 100 seconds on public cloud (HDD)
• Over 30-hop deep penetration in real-time

Supporting Products:

• Ultipa Manager v1.x
• Ultipa Transporter v1.x

Release Notes:

Ultipa Genesis, version 1.0 of Ultipa Graph System, introduces the revolutionary High-Density Graph Computing Engine. The engine leverages the Ultipa Index-free Adjacency data structure, ensuring swift and efficient processing of intricate graph data. Additionally, the Ultipa Dynamic Pruning system optimizes and prunes data in real-time during filtering and computation, which greatly enhances the engine performance.

The High-Density Graph Computing Engine boasts advanced capabilities like AB-path finding and K-Hop counting within microseconds, auto-spreading the neighborhood information of any data under concern and allowing rapid and deep traversal of large graphs with penetration up to 30 layers.

The two powerful products that come in bundle, Ultipa Transporter and Ultipa Manager, facilitate data import/export procedures and provide visualization of graph data queries and management.

Ultipa Manager v1.x offers 2D/3D visualization of graph query parameters and results, simplifying exploration and understanding of complex, high-dimensional graph data through a user-friendly web interface.

With Ultipa Transporter v1.x, loading massive datasets becomes effortless. For instance, the Twitter-2010 dataset comprising 1.5 billion nodes and edges can be loaded from local files into the remote database in just 100 seconds.

Ultipa Genesis heralds a paradigm shift in graph computing, delivering unmatched performance, scalability, and flexibility.