Changelog

RSS feed test entry

This is a test entry to verify the RSS-to-email pipeline. Will be removed after confirming Blogtrottr delivery works.

Static site & SPA hosting on Object Storage

You can now host static websites and single-page applications (React, Vue, Angular, Svelte) directly on IBEE Object Storage with the CDN in front for HTTPS, custom domains, and global edge caching.

What’s new

• End-to-end guide: build → bucket → CDN → custom domain
• SPA fallback (404 → index.html) for client-side routing
• Cache header recommendations for hashed assets vs index.html
• One-command redeploy via aws s3 sync

Flex shapes on Cloud VMs

Flex shapes let you pick exact vCPU and RAM amounts instead of choosing from fixed sizes. Useful for workloads that don’t fit standard shapes — pay only for what you provision.

Highlights

• Per-core RAM ratio: 1 GB to 16 GB per vCPU
• Vertical resize without VM rebuild
• Available in VGA, HYD1, and IAD

AMD EPYC bare metal servers in HYD1

AMD EPYC 9654 (Genoa) bare metal servers are now provisionable in the Hyderabad region. Single-tenant hardware with full hardware control and no hypervisor overhead.

Configuration

H100 GPU VMs now available in HYD1

NVIDIA H100 80GB GPU instances are now available in the Hyderabad (HYD1) region. Previously limited to IAD, you can now run training and inference workloads with sub-millisecond latency for India-based applications.

Available shapes

Layer-7 routing rules on Load Balancer (Beta)

Load Balancer now supports path-based and host-based routing rules in addition to round-robin. Route /api/* to one backend pool and /* to another, all within a single LB.

Beta scope

• HTTP and HTTPS listeners
• Path prefix and exact match
• Host header routing
• Up to 50 rules per listener

NVMe-backed Block Storage in HYD1

Block volumes in HYD1 are now backed by NVMe SSDs by default. Existing volumes are migrated transparently — no action required.

Performance improvements

Managed Kubernetes 1.30 is now generally available

Managed Kubernetes 1.30 is now GA across all regions. New clusters default to 1.30; existing clusters can upgrade in-place via the dashboard or CLI.

Highlights

• In-place node upgrades — rolling node-pool upgrades with zero pod evictions on PDB-protected workloads
• Sidecar containers beta is now stable
• Pod-level resource limits for fine-grained scheduling
• CSI inline volume snapshots for stateful workloads using IBEE Block Storage

PostgreSQL 16 on Managed DB

PostgreSQL 16 is now available when creating a new Managed DB cluster. Existing clusters can upgrade in-place from 14 or 15 with the dashboard upgrade wizard.

What’s notable in PG 16

• Logical replication from standbys
• SQL/JSON constructors and IS JSON predicate
• Faster vacuum on large tables
• Improved monitoring views

Cross-region snapshot copy

Snapshots can now be copied across regions for disaster recovery. Previously snapshots were region-local — to recover in another region, you had to manually re-create the workload.

How it works

• Right-click a snapshot in the dashboard → Copy to region → pick destination
• Or via CLI:

  $
  ibee snapshot copy <snapshot-id> --to-region IAD
• Background copy at line-rate within the IBEE backbone (no egress charges between regions)
• Destination snapshot is independent — deleting the source doesn’t affect it

Bulk import Firewall rules from CSV

Migrating from another cloud or consolidating rules across projects? You can now upload a CSV of firewall rules instead of clicking through each one.

CSV format

direction,protocol,port_range,source,action,description
ingress,tcp,22,10.0.0.0/8,allow,SSH from internal
ingress,tcp,443,0.0.0.0/0,allow,HTTPS public
egress,tcp,5432,10.1.0.0/16,allow,DB connections

Secret Manager v1 API is deprecated

The Secret Manager v1 API will be removed on 2026-09-30. Migrate to v2 before then to avoid disruption.

What’s different in v2

• Versioned secrets with rollback
• Per-field encryption metadata
• KMS-backed encryption keys (BYOK supported)
• Faster reads (~3× lower p99 latency)