メインコンテンツまでスキップ
バージョン: User Guides (Cloud)

January 2026 Release Notes

2026-01-14

Milvus v2.6.x new features​

  • Time-zone-aware timestamp support — Supports the TIMESTAMPTZ data type for storing, comparing, and filtering globally consistent timestamps—without manual time-zone handling. For details, refer to TIMESTAMPTZ Field.

  • Highlighter — Annotates matched terms with customizable tags and fragment-level context, making full-text search results easier to interpret and debug. For details, refer to Text Highlighter.

Dynamic Replica Autoscaling​

We are introducing Intelligent Replica Autoscaling, a key feature designed for high-QPS environments with fluctuating demand. It automatically adjusts your cluster’s replica count based on real-time traffic patterns.

  • Load-Adaptive Scaling: Automatically scales replicas up during periods of high traffic and scales them down during low demand, optimizing both performance and cost.

  • Zero-Touch Reliability: With simple resource guardrails, the system automatically handles unpredictable traffic spikes, ensuring consistent performance without manual intervention.

For details, refer to Scale Replica.

Advanced Scheduled Scaling with Cron​

We have upgraded our scheduling engine to orchestrate complex, predictable business cycles. You can now automate precise scaling strategies for both CUs and Replicas using industry-standard Cron expressions.

  • Flexible Scheduling Strategies: Move beyond basic daily schedules. Utilize standard Cron syntax (e.g., 0 9 * * *1-5) to define intricate rules, such as "scale up exclusively for month-end."

  • Multi-Schedule Logic: Configure independent, layered schedules for the same cluster, enabling you to adjust resource profiles for peak weekdays and off-peak weekends, optimizing efficiency in line with your business realities.

For details, refer to Scale Query CU and Scale Replica.

Global Cluster Private Preview​

We are excited to announce the Private Preview of the Global Cluster for the Zilliz Cloud Business Critical Plan.

The Global Cluster creates a unified database architecture across multiple geographic regions by linking a primary cluster with cross-region secondary clusters for automated replication. This solution provides robust Disaster Recovery (DR), ensuring your mission-critical applications remain resilient and your data durable, even in the event of a regional outage.

  • Automated Global Deployment: The system handles the seamless orchestration of Primary-Secondary topologies with one click, allowing you to provision a Global Cluster in a single step with automated data replication channels.

  • Seamless DR Expansion: Support for the dynamic addition of secondary clusters to active production instances. You can now upgrade a running dedicated cluster to a multi-region global architecture smoothly without service interruption or downtime.

  • Enhanced Observability: A new Global Topology dashboard provides a unified view of your cluster hierarchy. You can now monitor real-time replication latency and synchronization status across regions from one interface.

Coming Soon: We are expanding our resilience toolkit. The next phase will introduce Failover for automatic switching during region-level outages, and a Global Endpoint that reroutes SDK traffic, significantly reducing Recovery Time Objectives (RTO).

For details, refer to Global Cluster Explained, Create Global Cluster, and Manage Global Cluster.

BYOC - Full Autoscaling Suite aligns with SaaS​

Bring Your Own Cloud (BYOC) deployments now support the complete Zilliz Cloud autoscaling ecosystem. This update aligns BYOC with our SaaS offering, granting access to all previously released optimizations (like automatic scale-down) as well as the latest capabilities.

  • Dynamic Scaling: Available for both CUs and Replicas, the system intelligently adjusts resources based on real-time load to optimize performance and cost, with simple Min/Max configuration.

  • Scheduled Scaling: Full support for the new Advanced Mode. Users can now leverage standard Cron expressions and multi-schedule logic to automate precise resource adjustments for complex, predictable business cycles.

For details, refer to Scale Query CU and Scale Replica.

BYOC - Support & Troubleshooting Access Control​

Gain authority over operational access to your data plane. Ensuring that Zilliz engineers have access to your infrastructure only when explicitly permitted.

  • Just-in-Time (JIT) Privileges: Grant temporary access during troubleshooting windows and revoke it immediately once resolved.

  • Operational Isolation: Revoking access creates a strict barrier without disrupting essential observability pipelines (Metrics, Logs, and Alerts).

  • Governance & Compliance: All access grants and revocations are immutably logged in your Audit Logs for complete accountability and security reviews.

For details, refer to Deploy BYOC on AWS, Deploy BYOC-I on AWS, and Deploy BYOC on GCP.

Enhancements​

  • Collection TTL and AutoID Settings: You can now monitor and modify the collection TTL and Allow insert AutoID settings directly from the Collection Overview GUI. For details, refer to Set Collection TTL and Modify Collection.

  • Data Import: Support for JSON lines format (.JSONL and .NDJSON extensions) is now available. For details, refer to Import from a JSON/JSON Lines File.

  • Milvus Endpoint Migration: Now supports Geometry and Struct data types, enabling seamless migration of collections with spatial shapes and deeply nested attributes.

  • Job Details View: The side drawer UI has been refreshed for improved navigation and better user experience.

  • BYOC - S3 Bucket Support: You can now deploy clusters with dedicated S3 buckets, providing granular data isolation and independent lifecycle management.

  • BYOC - AWS KMS Integration: AWS KMS (CMEK) integration for S3 bucket encryption has been added, satisfying strict security compliance standards.

  • Enhanced Metrics Dashboards: Visual threshold guidelines have been added to help users identify optimal utilization levels for scaling CUs and Replicas.

  • RESTful API & Terraform Enhancements: Now supports Auto Scaling, Cross-Region Backup, Tiered Storage for Create Cluster, and Business Critical Plan for Create Project, improving disaster recovery and storage management, enabling more efficient automation programming.