Zens Distribution

SmartX

Full-stack hyperconverged infrastructure with native virtualisation, distributed storage, networking and Kubernetes.

Request a quoteSee where it fits in the solution stack
SmartX logo
Category
Infrastructure
Distributed in
Hong Kong, Macau
Vendor website
smartx.com

Overview

SmartX, founded in 2014, makes the Enterprise Cloud Platform (ECP). It combines native virtualisation (ELF), distributed block storage (ZBS), software-defined networking (Everoute), Kubernetes services, data protection and central management (CloudTower) in one platform, and SmartX positions it as an alternative to VMware.

ECP 6.3 and the SMTX AI Platform are the current releases.

ZENS is the official SmartX distributor in Hong Kong.

SmartX reports that IDC ranked it first in China’s HCI software market for 11 consecutive quarters (Q1 2023 to Q3 2025).

Platform components

  • SMTX OS

    Native ELF hypervisor with ZBS distributed storage. Compatible with VMware ESXi.

  • CloudTower

    Central management, intelligent operations and multi-cluster control.

  • Everoute

    Software-defined networking, micro-segmentation, load balancing, VPC and zero-trust networking.

  • Backup, DR and Kubernetes

    Synchronous and asynchronous replication, active-active disaster recovery (RPO=0) and a native container platform.

ECP 6.3 performance

Results SmartX reports for a 3-node cluster.

  • Storage I/O

    4K random read of 11M+ IOPS, sequential read of 130+ GiB/s and 4K random read latency under 100 µs.

  • What changed from v6.2

    Intel DSA hardware acceleration, the IO_uring async framework, RDMA multi-link aggregation and multi-storage-instance parallelism. SmartX reports IOPS up 360% and throughput up 260% against v6.2.

  • High availability

    HA for the RDMA storage network (cross-NIC bonding), for VMs using SR-IOV NICs and for vGPU workloads, plus NIC teaming.

Disaster recovery and security

  • Synchronous replication

    VM-level replication with RPO=0, with failover and DR drills.

  • Active-active DR

    Stretched clusters across sites from four nodes, with placement group policies and automatic management-plane failover.

  • Security and compliance

    AES-256 data-at-rest encryption, agentless antivirus, national cryptographic algorithm encryption, a built-in KMS, live migration encryption and ERSPAN network auditing.

Editions

All editions run on the same SMTX OS foundation.

  • ECP Essential

    Core HCI with full virtualisation (ELF), distributed block storage with erasure coding, encryption at rest, agentless VM backup and CloudTower management. Replication is an add-on. Best for small deployments, branch offices, ROBO and dev/test.

  • ECP Standard

    Adds SMTX Kubernetes Service (SKS), file storage for external hosts, built-in async and sync replication, Pod-VM integrated networking and network observability. Firewall and load balancing are add-ons. Best for mid-size enterprises, VDI, containerised apps and general production.

  • ECP Advanced

    Adds built-in Everoute SDN with micro-segmentation and load balancing, active-active stretched clusters, Kubernetes on bare metal and CloudTower HA. VPC networking is an add-on. Best for finance, healthcare, core databases, multi-site DR and mission-critical production.

SMTX AI Platform

On-premise generative AI infrastructure built on ECP, delivering models as a service in production.

  • Model management

    Text generation, embedding and reranking models. Import from Hugging Face or upload your own, from a catalogue with predefined inference engines and resource specs.

  • Compute pool

    Heterogeneous GPU pooling across NVIDIA and AMD, on bare metal, VMs and Kubernetes, with GPU sharing and isolation.

  • Inference

    KVCache-aware multi-replica scheduling and load balancing for high-concurrency inference.

  • Multi-tenant governance

    Tenant isolation, fine-grained RBAC, and API key management with token quotas and usage statistics.

  • Monitoring

    From GPU status up to inference services, including API token usage, request throughput and model performance, in one dashboard.

  • Deployment

    On-premise on ECP, on bare-metal GPU servers, or a hybrid of both in one pool.

  • Ecosystem partners

    Joint validations with Baihai IDP, Fit2Cloud MaxKB, DeepSeek and Hugging Face.

Specifications

Maximum hosts per cluster
255
Maximum raw capacity per node
256 TB

Request a quote for SmartX

Tell us about the project and we will come back with a quotation.

Request a quote

Or call +852 3696-1050