Performance That Scales
Real benchmarks against real competition. Every number independently reproducible.
Test Environment
Both systems tested on identical hardware: 8 vCPUs, 16 GB RAM. 50 queries per thread across 12 concurrency stages (1 to 128 threads). Zero failures across all test runs. March 2026.
Standard MDX SELECT
Large result-set retrieval across profit centers, time periods, and scenarios.
TOPCOUNT Analytical Query
Top-20 ranking over a large member set. Measures analytical computation under concurrency.
Why the Architecture Matters
Ardello uses persistent WebSocket connections, one socket per client, kept alive for the session. SSAS uses ADOMD.NET over TCP with a new connection per query. At low concurrency, the difference is modest. At 128 threads, connection management overhead dominates SSAS response times. This is not a tuning problem; it is an architectural ceiling.
Degradation Under Load
The most revealing metric is not absolute speed; it is how gracefully a system degrades as concurrency increases.
Effective Throughput
Ardello achieves 36.6x throughput scaling on Test A compared to SSAS's 4.0x, a 9x advantage in effective throughput. Even with the more demanding TOPCOUNT query, Ardello delivers 13.2x throughput scaling vs SSAS's 3.5x.
Where Ardello Fits
Every existing option solves only part of the problem. Legacy OLAP engines like SSAS are genuinely multidimensional but stay tied to on-premise infrastructure. Cloud columnar engines like ClickHouse scale effortlessly but speak only SQL. Semantic layers like AtScale exist to patch the limitations of SQL-based solutions, layering multidimensional modeling on top of a warehouse without an engine of their own. BI tools like Power BI ship polished reports but model data in flat tables, not true cubes. Ardello is the only platform that unites a native multidimensional MDX engine with cloud-native, browser-direct delivery.
| Capability | Ardello | SSAS | ClickHouse | AtScale | Power BI |
|---|---|---|---|---|---|
| Native MDX Engine | Translates | DAX-first | |||
| True Multidimensional Modeling | Virtual | ||||
| Near-Instant Round-Trip | Varies | DW-bound | |||
| On-the-fly DB Provisioning | Managed | ||||
| WebSocket Delivery | |||||
| Dev / Test / Prod | Premium | ||||
| Integrated Report Builder | |||||
| Model / Report Separation | N/A | N/A | Partial |
Test: 50 queries per thread × 12 concurrency stages (1–128 threads). 8 vCPUs, 16 GB RAM. Zero failures. March 2026.
Ready to See It in Action?
Get a technical walkthrough of the Ardello platform. See real benchmarks on your data.
Get in Touch