Platform / Performance

The speed of local NVMe. The durability of cloud storage

Tessell runs every database on patented NVMe-backed infrastructure with up to 2 million IOPS, no metering and no throttling. Your most demanding workloads stay fast under load.

Talk to a Specialist

Up to 2 million IOPS

on a single database, NVMe-backed configurations

No IOPS

metering

Up to 100 Gbps

networking

The Tessell difference

Speed built into the architecture, not billed by the tier.

IOPS metering turns speed into a line item, and heavy workloads hit a ceiling. Tessell removes the meter. Performance comes from patented, durable NVMe architecture, with no throttling and no per-tier billing.

No metering, no throttling

IOPS are never capped or billed by tier. Workloads run at full speed under sustained, concurrent load.

Pay for the shape, not the throughput

Performance scales with the compute shape you choose, not a separate IOPS bill, so cost stays predictable as you grow.

Proven, not asserted

Benchmark results you can reproduce in your own account, with the shapes and workloads shared.

Durable by design

Local NVMe speed made persistent by Tessell's architecture, with zero data loss and multi-AZ recovery.

The foundation

Built on Tessell's patented durable NVMe architecture

Performance starts with locally attached NVMe, the fastest storage in the cloud, made durable and persistent by Tessell's patented architecture with zero data loss. Up to 2 million IOPS, multi-AZ, with near-zero RTO.

2M IOPS on NVMe-backed configs
Multi-AZ, near-zero RTO
Zero data loss

Customer results

Faster where it counts

database query performance at a large investment firm

faster month-end reporting

transaction volume at the same cost

Benchmarks

Measured, repeatable, verifiable

Tessell shares benchmark results so your team can validate performance against your own workloads, engine by engine.

Oracle

165,569 sustained IOPS · 8 vCPU high-performance shape SLOB benchmark on NVMe-backed shapes for demanding Oracle workloads.

SQL Server

Strong transactional throughput HammerDB · NVMe-backed shapes Strong transactional performance under a standard HammerDB workload.

PostgreSQL

165,521 sustained IOPS · 8 vCPU high-performance shape Standard PGIO workload with no IOPS throttling.

MySQL

High, consistent throughput Sysbench · NVMe-backed shapes Sustained transactional throughput under a standard Sysbench workload.

Put it to the test on your own workload.

We will size the right high-performance shape for your workload and run the benchmarks with you.

Talk to a specialist

Configurations and scale

Your workload. Your shape. Your cloud.

Choose the compute shape and cloud that fit each workload. High-performance NVMe shapes are available where your databases already run. Choose your:

Compute shape
Storage capacity
Cloud and region
Engine

Book a demo

How Tessell sustains performance

Fast on day one, and every day after

Consistent under load

Performance holds under sustained, concurrent workloads, not just in short bursts.

Full observability

Built-in performance monitoring surfaces query and I/O behavior so you can tune with confidence.

Availability Machine

Continuous protection runs alongside performance, with no trade-off between speed and safety.

Dedicated team

A dedicated engineering team, accountable from POC to production.

Fortune 500 companies, global banks, and government agencies trust Tessell with their most critical databases

Frequently Asked
Questions

Through patented architecture that pairs locally attached NVMe, the fastest storage in the cloud, with a durability layer that makes it persistent. You get the speed of local disks with the resilience of cloud storage.

No. IOPS are never metered, throttled, or billed by tier. You pay for the compute shape you choose, and performance scales with it.

You pay for the shape, not the throughput. There is no separate IOPS charge, so cost stays predictable as the workload grows.

At provisioning you pick a High Performance (NVMe) or Standard compute shape per service, so you right-size each workload instead of over-provisioning across the board.

Oracle, SQL Server, PostgreSQL, and MySQL, on AWS and Azure today, across transactional and analytical workloads that demand high, sustained IOPS.

A specialist sizes it with you from your current IOPS, throughput, and latency profile, so the shape matches the workload.

Yes. Tessell shares the shapes, workloads, and configuration so your team can run the same tests in your own account against your own data.

Yes. Tessell's architecture makes NVMe durable and persistent with zero data loss, backed by multi-AZ availability and near-zero RTO.

Talk to a specialist. We will walk your team through the architecture, share our reports, and answer your questions.