MagicCube Data Portal — Product Guide #

Audience: Data analysts, business users, data administrators | Updated: August 2026

Table of Contents #

1. Product Overview
2. What Can I Do
3. System Entry & Roles
4. User Portal Guide
5. Admin Portal Guide
6. FAQ
7. Glossary

1. Product Overview #

MagicCube Data Portal is a unified data asset management and self-service data query platform. It centralizes scattered data tables, reports, metrics, and API interfaces across the enterprise, enabling business users to easily find data, understand data, and use data.

Think of it as: if data is a library, MagicCube is the catalog system + borrowing system + recommendation system.

2. What Can I Do #

3. System Entry & Roles #

3.1 Page Entry Points #

3.2 Roles & Permissions #

Portal uses OA permission codes for access control, with Django Group as fallback.

4. User Portal Guide #

Access: /portal/

4.1 Homepage Overview #

4.2 Data Map — Find Data #

The Data Map is the core entry for finding data, supporting multi-condition search and directory browsing. Search scope covers data tables, metrics, reports, datasets, and APIs.

Search methods:

1. Enter keywords (fuzzy search)
2. Filter by: data layer (ODS/DWD/DWS/ADS/DIM), business domain, update frequency, security level, owner, status, tags
3. Switch type: All / Tables only / Metrics only / Reports / Datasets / APIs
4. Switch view: Card / Table

Permission status icons:

• 🟢 Green = Authorized (can use in self-service analysis)
• 🟡 Yellow = Pending approval
• 🔒 Gray = Not authorized (click “Apply for Permission”)

4.3 Self-Service Analysis — Query Data #

Access: /portal/self/

Visual drag-and-drop query — no SQL needed. Drag fields into dimension, measure, filter, and sort areas.

Config areas:

Result features:

• Table view with pagination, sorting, field selection
• Chart view (line, bar, etc.)
• Pivot table support
• Local Analysis Mode: Auto-loads up to 5000 rows for client-side processing
• Data processing: text, numeric, date transformations, ranking, formatting
• Save/load queries (export/import as JSON)
• Export CSV / Excel
• Forecast analysis: Time series prediction with confidence intervals (MAPE, R²)

4.4 My Applications #

View all submitted applications and metric requests:

• Permission applications: View status (Authorized/Pending/Rejected)
• Metric requests: Submit new metric needs with business background, calculation logic, priority

4.5 Dataset Builder — Make Data #

Access: /portal/design/

Visual drag-and-drop dataset design — join multiple tables, select fields, configure aggregations, and publish as reusable datasets.

Design flow:

1. Drag tables from authorized directory tree to canvas
2. Auto-detect JOIN conditions (same-name fields, foreign key patterns)
3. Select output fields, set aliases
4. Configure aggregation functions (SUM/COUNT/COUNT_DISTINCT/AVG/MAX/MIN)
5. Add calculated fields (custom SQL expressions)
6. Configure filter conditions (12 operators)
7. Preview SQL and data
8. Publish as dataset asset

Supports multi-step CTE (Common Table Expression) mode for complex data transformation pipelines.

5. Admin Portal Guide #

Access: /portal/admin/

5.1 Asset Management #

Core workspace for maintaining data assets. Left: asset directory tree. Right: overview dashboard and asset list.

Asset types (4 tabs):

• Data Tables: Edit business properties, tag, sync metadata, preview data, configure quality rules
• Reports: BI reports, charts, Excel reports
• Datasets: File, API, database, and visually-built datasets
• APIs: External API interfaces

Key features:

• Data dashboard (table count, layer distribution, metric count, quality score)
• Directory tree by business domain
• Edit table attributes (domain, owner, update frequency, security level, description)
• Tag management (multi-tag, custom colors, batch operations)
• Metadata sync (single/batch, sync logs, change logs)
• Field maintenance (comments, active status, manual add/remove)

5.2 Data Lineage #

Visual representation of dependencies between tables and metrics — understand where data comes from and where it goes.

Operations: Expand upstream/downstream, search, view details, pan/zoom canvas, reset view.

Node colors: Blue=atomic metric, Green=derived metric, Gray=ODS, Orange=DWD, Blue=DWS, Purple=ADS, Cyan=dimension table.

Detail panel: Shows type, SQL, upstream/downstream, lifecycle status, field list, quality score.

5.3 Metric Center #

Access: /portal/index/

Unified platform for all business metrics — definition, lifecycle, dimension management, and business qualifiers.

Four tabs: Metric Library, Metric Definition, Dimension Management, Business Qualifier Management.

Three metric types:

• Atomic: Basic measure (e.g., Order Amount = SUM(amount))
• Derived: Based on atomic with conditions/time aggregation (e.g., Monthly New Customer GMV)
• Composite: Multiple metrics combined (e.g., Average Order Value = Amount / Order Count)

Metric lifecycle: Draft → Apply for Online → Under Review → Online → Change/Offline

Auto metric generation: System recommends metrics based on table fields.

5.4 Data Quality #

Access: /portal/quality/

Quality dashboard with rule count, pass/fail distribution, 7-day trend, and failure Top 10.

10 quality rule types:

• Null check, Uniqueness check, Value range, Enum check, Regex validation, Length check, Consistency check (cross-table SQL), Timeliness check, Data volume fluctuation, Custom SQL

Quality scoring: 0-100 per table. Severe rules deduct 30 pts, general 15 pts, info 5 pts. Auto-updated after each check.

5.5 Data Security Center #

Access: /portal/security/

Field-level masking (20+ strategies): Hide, phone mask (138****1234), email mask, ID card mask, bank card mask, address mask, name mask, salary mask, date blur, SHA256 hash, MD5 hash, etc.

System auto-recommends masking strategies based on field names and data types.

Row-level filtering: Control which data rows users can see. Supports 12 operators (=, !=, >, <, >=, <=, in, not_in, like, between, is_null, is_not_null).

Multiple conditions within a rule use AND; multiple rules on same table use OR. User-specific rules override default rules.

5.6 Approval Management #

Handles two types: data permission applications and metric lifecycle applications.

Flow: View pending → Check details → Select records → Batch approve/reject → Add review note → Confirm.

5.7 Usage Analysis #

Access: /portal/usage/

Query statistics: total queries, unique users, total rows, average latency. Trends by day, Top users, Top tables, latency distribution. Recent query details with SQL preview. Asset usage monitoring (views, favorites, queries per table).

5.8 Data Service & API Management #

Managed via Asset Management → API tab. Configure API name, associated table, query SQL, return fields, access permissions. Masking rules from Data Security Center apply to API responses.

5.9 Materialized Views #

For accelerating common queries, especially with StarRocks.

States: Active, Inactive, Building, Failed.

Refresh strategies: Manual, Auto, On-commit, On-demand.

5.10 System Settings #

6. FAQ #

Q1: Can’t find a table in Data Map?

• Table not registered → Contact admin to add
• No permission → Apply for access
• Keyword not precise enough → Try partial keywords

Q2: Can’t see a table in self-service analysis? Only authorized tables appear. Check permission status in Data Map first.

Q3: How long until metric request gets a response? High priority: 1-3 business days; Medium: 3-5 days; Low: batched by schedule.

Q4: How to export self-service analysis results? Click “Export CSV” or “Export Excel” above the result area. Export volume depends on result mode (current page for server-side pagination, up to 5000 rows for local analysis mode).

Q5: User portal vs admin portal URLs?

Q6: Why can’t I see certain fields in self-service analysis? Field hidden by masking rule, partial table permission, inactive field, not in dataset schema, or filtered out by row-level security.

Q7: How to configure data masking? Admin goes to Data Security Center (/portal/security/), selects table, views fields, confirms or modifies auto-recommended masking strategies, saves.

7. Glossary #

For questions, please contact the data management team.