How we measure our performance figures
Figures on this site such as 0.95 ms query time or up to 70% disk savings are internal test results. This page explains under which conditions each figure was measured, what it does and does not cover, and how you can verify it in your own environment.
Which number, under which conditions?
Each row states one claim, its measurement context and its limits. We do not put any figure into marketing copy without its context.
| Claim | Value | Measurement context | Limits and notes |
|---|---|---|---|
| Historical query time | 0.95 ms · average | Internal test; measured on a server with 4 cores, 8 GB RAM and 100 GB disk, on a firewall log set of 2 billion records, with ingestion running at 2,500 EPS. No indexing database; record age does not change the result. | This is an average; the p95 and query-type breakdown are provided in the PoC report. Tests have been repeated regularly since 2021; last measurement November 1, 2026. Network latency and concurrent users affect results. |
| Archive compression | Down to 1/40 | Internal test; the patented Minimize algorithm compressing archive data held on disk. | Roughly 40 GB of raw logs takes up 1 GB on disk. Because data is stored on disk for years rather than in RAM, compression is part of the architecture. The ratio varies with data type. |
| Disk savings (deduplication) | Up to 70% | Upper value measured in an internal test scenario; Deduplication strips repeating records | The ratio varies with log source diversity: high for uniform firewall traffic, lower for diverse application logs. |
| Memory and CPU consumption | 50% / 75% less | Internal test; comparison against a representative Elasticsearch/JVM-based stack on the same data set | The comparison is representative; a branded competitor measurement is only done in a PoC on identical data and hardware. |
| Single-server EPS capacity | 50,000 EPS | OrianaLOG, single server, internal test; sustained ingestion with zero event loss | Reference values including SIEM correlation load are in the sizing table. |
| Peak EPS reference | 100,000 EPS | 5-server Kubernetes cluster reference configuration | Peak value; for sustained load, use the nominal EPS in the table. |
| Archive disk ratio | ~0.15× raw volume | Derived from reference table values: 5,000 EPS · 365 days → ~73 TB raw, ~11 TB archive | Deduplication + compression combined; the log cost calculator uses this ratio. |
| Installation time | 1 day | Standard single-server installation; field experience | Multi-server clusters and custom integrations extend the timeline; it is finalized in the PoC plan. |
Our measurement principles
Same data set
Every comparative measurement uses the same log data set, the same retention policy and the same query scenario. Results obtained on different data sets are never placed side by side.
Same hardware
Cross-product comparison is only meaningful on equivalent CPU, RAM and disk. We never publish a comparison figure without stating the hardware.
Realistic load profile
High daytime EPS and nighttime archiving/compression load are tested together; "showroom" measurements taken on an idle system are not used.
The "representative" label
In every comparison where we did not measure a named competitor product, the other side is labeled "representative"; definitive comparison happens only in your own PoC environment.
The most reliable benchmark is the one run on your own data.
Internal test results point the way; they are not enough for a decision. In a 1–2 week PoC we connect your own log sources, take the measurements together and hand over the results as a report you can present to management.
We write the measurement plan, success criteria and acceptance thresholds together, before the PoC starts.
What we measure together in the PoC
- Query response time: on fresh and 6+ month old data, p50/p95
- Sustained EPS capacity and event loss at peak load (target: 0)
- Deduplication ratio and actual disk growth
- CPU / RAM consumption under the day/night load profile
- Report generation time and source onboarding time
Measurement results are valid only for the environment they were measured in; we do not promise "the same result at any scale".
Let's write the measurement plan together
We run the tests in your environment, on your own logs. The resulting report stays with you, ready to take to management.