What’s New in TDH 5.2.1 & 5.2.2? A Deep Dive into Big Data Platform Enhancements
The article details the feature upgrades, performance improvements, bug fixes, and new capabilities introduced in TDH versions 5.2.1 and 5.2.2, covering components such as Inceptor, Slipstream, Rubik, Workflow, Manager, Guardian, and future roadmap for the big‑data platform.
TDH 5.2.2 Product Update
TDH 5.2.2 adds CentOS 7.5 support, enhances Inceptor partition tables and Compact functionality for DECIMAL fields, and improves the usability of Studio tools.
Inceptor
Fixed an error that occurred during compaction of ORC transaction tables containing DECIMAL fields.
Resolved the inability to use DECIMAL constants in ALTER TABLE ADD/DROP PARTITION statements. Example:
alter table test_partition drop partition(p=2.1);Slipstream
Fixed an issue where writing to RabbitMQ could leak data from the previous record when the previous record was longer than the current one.
Transporter
Changed the representation of non‑printable text delimiters from “%+hex” to “\
” (e.g., "\001", "\002") to align with Inceptor.
Fixed data misalignment when source tables contain NULL values.
Added support for Blob (binary large object) fields.
Rubik
Corrected the display of detection details in the instance management page.
Optimized permission‑information retrieval for faster operations.
Enabled printing of REST API parameters for easier interface analysis.
Workflow
Enabled online lifecycle management of published workflows without requiring offline steps.
Redesigned the Cron expression UI, removing the “week” field and providing a visual configuration interface.
TDH 5.2.1 Product Update
TDH 5.2.1 improves container networking with Flannel, adds UI‑driven installation of Flannel, kube‑dns, and kube‑cni, and enhances Manager’s loading performance. Guardian now fully supports Access Token authentication for web services.
Manager
Provides a UI entry for global Flannel service installation and configuration.
Allows monthly inspection report viewing.
Installation and upgrade improvements include Kerberos support on SUSE12, added steps for role addition and custom configuration, and HA‑aware Manager upgrades.
Performance optimizations speed up page loading, configuration file generation for large clusters, and hotspot map rendering.
Guardian
All web service components now support Access Token authentication.
Introduced a heartbeat mechanism to detect offline services and automatically remove them.
Inceptor Enhancements
Added new and extended UDFs, including LIKE ANY, LIKE ALL, and enhanced REGEXP_REPLACE with position and occurrence parameters. Example:
SELECT * FROM user_table WHERE first_name NOT LIKE ANY ('root%','user%');Extended DISTINCT support in UDAFs, window functions, and SELECT statements with GROUP BY rollup/cube.
Enabled push‑down of outer LIMIT into subqueries for more efficient execution.
Holodesk
Introduced Holodesk Meta Cleaner to remove invalid files from ZooKeeper and HDFS.
Supported Global Index usage in Shiva‑based clusters.
Performance Optimizations
Dynamic partition insert now auto‑matches and converts implicit type differences.
For partitioned non‑bucket tables, distribute by rand() reduces small file generation.
Static single‑value partition inserts now validate partition values beforehand.
Floating Scale feature mitigates overflow in DECIMAL calculations by discarding excess fractional precision; controlled by inceptor.floating.scale.decimal.
Slipstream
From version 5.2.1, Window Stream supports Kafka, RabbitMQ, socket, and custom data sources.
Search
Improved data storage format, boosting serialization/deserialization performance by ~30% and reducing storage by ~20%.
Optimized SELECT COUNT(*) push‑down, achieving 3–4× speedup on large datasets.
Discover
Fixed Notebook startup failure when Kerberos is enabled in version 5.2.
Transporter
Implemented cross‑system data flow import/export, exporting data streams in a compatible format for seamless upgrades.
Workflow
Added periodic cleanup of Transporter history records to prevent database bloat.
Rubik
Added “filter condition” option for two‑table joins in Cube and snowflake dimension pages.
Future Versions
TDH 6.0, expected mid‑September, will introduce Inceptor HA with load balancing and intelligent routing, real‑time OLAP cube tools, a visual stream task management UI, and enhanced stability and capacity for the Search engine.
StarRing Big Data Open Lab
Focused on big data technology research, exploring the Big Data era | [email protected]
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.
