top of page

Cloud Storage Costs 70% Less: Zettalane's Business Case for Hybrid Architecture

  • Writer: ctsmithiii
    ctsmithiii
  • 1 hour ago
  • 4 min read

Zettalane Systems cuts cloud storage costs 70% with hybrid ZFS architecture. Multi-cloud support across AWS, GCP, and Azure with two-minute deployment.



The Cloud Storage Cost Problem

Cloud file storage pricing creates real business constraints. A 100TB file system on AWS EFS costs $360,000 per year at $0.30 per GB monthly. Block storage costs less ($96,000 annually on EBS gp3 at $0.08/GB/month) but doesn't provide the shared file access that applications require.


Beyond pricing, performance limitations force businesses to over-provision capacity. Most managed file services only deliver advertised throughput at 100 TB or higher. Smaller deployments get throttled performance even when paying premium prices.


Zettalane's Approach

At the 66th IT Press Tour in January 2026, Zettalane Systems founder Supramani (Sam) Sammandam presented the company's cloud-native storage solutions. Zettalane offers 70% cost savings versus traditional cloud storage through a hybrid architecture that combines NVMe SSDs with object storage.


The company offers two products:

MayaNAS provides high-throughput NAS (NFS/SMB) using hybrid ZFS architecture. Metadata stays on fast NVMe drives while bulk data goes to object storage (S3, Google Cloud Storage, Azure Blob). Performance reaches 4GB/s throughput per node.

MayaScale delivers ultra-low latency block storage using local NVMe SSDs with NVMe-over-Fabrics. It provides 2.3 million IOPS with sub-millisecond latency.

Target Market

Zettalane focuses on small-to-medium businesses and developers who need cloud storage but face barriers with enterprise solutions. Most enterprise-managed storage services require 100TB minimum commitments. Zettalane supports deployments from terabytes to petabytes with no minimum capacity requirements.

The company currently has a team of about 10 people. Sam founded Zettalane in 2018 after 25 years in systems software engineering at companies including Dell, IBM, HP, and Lucent Technologies. He previously created MAYASTOR, an early software-defined storage platform, in 2007.

Business Model

Zettalane uses simple per-vCPU-hour pricing with no per-GB or per-IOPS charges. Cloud infrastructure costs (compute instances, object storage) are billed separately by the cloud provider.

The company reaches customers through three channels:

  1. Cloud Marketplaces—Self-service deployment through AWS, GCP, and Azure marketplaces with consumption-based billing

  2. Direct Sales - Enterprise accounts with technical evaluation and custom deployments

  3. Channel Partners - System integrators and managed service providers

All three products are live on AWS, GCP, and Azure marketplaces now.

Use Cases

MayaNAS Applications:

  • AI/ML training datasets with large sequential files

  • Media and entertainment video editing and rendering

  • Data analytics, including data lakes and ETL pipelines

  • Software development build artifacts and containers

  • Backup and archive with cost-effective object storage

MayaScale Applications:

  • High-performance databases (PostgreSQL, MySQL, Oracle)

  • Real-time analytics (ClickHouse, Druid, TimescaleDB)

  • Kubernetes persistent volumes

  • OpenZFS NFS on clouds without AWS FSX

  • CI/CD build systems requiring fast local storage

Cloud Partner Strategy

Zettalane focuses primarily on Google Cloud Platform, though it supports all three major clouds. GCP lacks some storage options available on AWS (like FSX) and provides generous network bandwidth allocations.

GCP allows Tier 1 premium networking for free on standard instances with 32+ cores. This provides 50-75 Gbps network bandwidth compared to the standard 32 Gbps. AWS and Azure charge extra for equivalent network performance.

GCP's older Y2 family instances include local NVMe storage at lower cost than newer Titanium SSD instances. For customers running ephemeral workloads, this represents significant savings.

Zettalane holds partner status with all three providers:

  • GCP Silver Partner

  • AWS Partner (Partner Network)

  • Azure Partner (Microsoft Partner Program)

Deployment Time

Both products deploy via Terraform infrastructure-as-code in approximately two minutes. The Terraform modules handle all cloud-specific configuration, including networking, storage provisioning, and high availability setup.

This deployment speed is important for organizations that need storage quickly for time-sensitive projects or want to test the solution before committing.

Data Sovereignty

The software runs entirely in the customer's private VPC with no call-home requirements. Zettalane doesn't monitor deployments, collect telemetry, or require internet connectivity. This addresses data sovereignty concerns, particularly relevant in European markets.

There are growing discussions in Europe about cloud sovereignty and reducing dependence on American cloud providers. Zettalane's architecture could support on-premises deployments, though the company currently focuses only on public cloud.

Competitive Positioning

Zettalane competes against:

  • Cloud provider native services (AWS EFS, FSX, Google Filestore)

  • Enterprise cloud storage vendors (Silk, volumes.io)

  • Traditional storage vendors running on cloud

The company differentiates through:

  • Lower cost (70% savings claimed)

  • Server-side replication architecture that reduces client overhead

  • Support for both block and file protocols in MayaScale

  • NVMe-over-RDMA capability (competitors typically support only NVMe-TCP)

  • Simple per-vCPU pricing with no surprise bills

  • No minimum capacity requirements

Performance Validation

Zettalane provides benchmark scripts on GitHub that customers can run to validate performance claims. This transparency approach builds credibility, particularly important for a newer company without established market presence.

The company uses open-source components (OpenZFS, Linux MD RAID) for the data plane, which reduces perceived risk. Zettalane's proprietary code handles only the control plane.

Current Status

The products launched on cloud marketplaces in 2024-2025. The company is building its initial customer base and working to establish market awareness.

Zettalane presented at the OpenZFS Developer Summit to validate its approach with the ZFS community. The IT Press Tour presentation represents another step in building market visibility.

Growth Path

The roadmap includes:

  • Kubernetes CSI driver to address the container persistent storage market

  • Cloud-native Lustre support for HPC workloads

  • pNFS FlexFiles for scale-out NFS deployments

These additions would expand addressable use cases while maintaining the core cost and performance advantages.

Investment Required

Organizations can test Zettalane with minimal investment. The software runs on pay-as-you-go cloud infrastructure with hourly billing. No long-term commitments or large upfront purchases are required.

This aligns with the target market of smaller organizations that need flexibility and want to avoid expensive proof-of-concept projects.

Questions About Scale

The presentation didn't address how the solution performs at enterprise scale (multi-petabyte deployments) or whether the architecture has limitations at considerable scale. The active-active HA uses two-node pairs rather than larger clusters, which might limit some enterprise use cases.

Partnership opportunities exist with companies offering complementary capabilities (the presentation mentioned potential partnerships with AmieSpace for pNFS and Lustre-related vendors).


Bottom Line

Zettalane addresses a real pain point: cloud storage costs too much and forces small organizations to over-provision capacity. The hybrid architecture approach makes sense technically. The company now needs to prove it can execute on sales and support at scale.

The 70% cost savings claim is significant if validated. Organizations spending six figures on cloud storage annually should evaluate whether Zettalane's approach fits their use cases.

 
 
 

© 2025 by Tom Smith

bottom of page