Light sport aviation has matured by writing things down: consensus standards, maintenance records, pilot logbooks, inspection schedules. What began as an informal community grew up when people agreed on shared formats and kept their evidence in order. Security data is travelling the same road, and the practice called SecDataOps is where that effort now gathers. This state-of-the-field piece reviews where it stands in late 2026, what has settled, what is still moving and what a prudent operator should do about it.
Most of what follows rests on the Ops Atlas, and its page covering the SecDataOps discipline was last checked in October 2026. Where I describe movements in standards or products, they come from that entry, and any numbers belong to it.
A label still finding its feet
The first thing to say is that the name is young. The Atlas describes it as an emerging, supplier-led label, in public use since at least March 2023, with no standards body behind it. A few companies promote it, a podcast series on the theme has run since 2024, and an industry analyst firm gave the underlying tool category the less catchy name of data pipeline management in late 2024. Meanwhile, governments continue to publish the same guidance under older titles, with a 2023 American draft, a 2021 federal memorandum and Australian-led guidance in 2024 all describing log management and event logging.
The editorial point the Atlas makes is useful for readers sorting through brochures: SecDataOps builds data for security, while a sibling discipline secures business data. If a product claims both, ask which job it is actually doing.
What has settled
Schemas are converging
Shared event models have moved from aspiration to practice. The open cybersecurity schema reached version 1.9.0 on 3 August 2026 under the Linux Foundation, and its event type identifier is calculated from a class identifier and an activity identifier. A separate common schema, now at 9.5.0, is converging with OpenTelemetry. Several vendors still maintain their own, so mapping between models remains normal work. What has settled is the expectation that a sign-in from two products should look the same after normalisation.
Raw data first, tidy data second
In the Atlas view, established practice is to store the original lines on low-cost media and derive the tidy version downstream, since parsers slip and open schemas withdraw attributes from one release to the next. That view has shifted from advice to default.
Retention is being argued on evidence
The conversation has moved away from a stock three-month default. The Atlas summarises a 2026 review of the prior year’s incident work. Intruders typically lingered two weeks worldwide, espionage and covert-employee cases lasted around four months, and one persistent implant survived for about 400 days. The review’s authors say that routine quarterly retention hides the initial point of entry. Regulators have also set explicit tiers, including a US federal scheme with active and cold periods and an Indian 180-day rule dating from 2022.
What is still moving
Tooling
The open-source collector the Atlas highlights shipped version 0.59.0 on 6 October 2026, and its documentation still warns about strict parsing behaviours and non-durable memory buffers. A redaction processor in the OpenTelemetry collector remains alpha for logs, with a keyed hash recommended over an unkeyed one. Meanwhile, one public toolkit that many small teams relied on lost official support in May 2026, a reminder that free tooling carries a maintenance risk.
Platforms and tiers
Cloud vendors are building cheaper storage options for security data. The Atlas mentions a data lake from one cloud provider that stores records as Parquet in the open schema, though it did not support object lock as of October 2026, and a security-analytics lake from another with retention of up to twelve years. A third offers basic and auxiliary plans in its monitoring service: the auxiliary plan has minimal ingest cost, slow queries and no alerts. Where this is heading seems plain even though details shift every quarter: inexpensive raw storage, topped by a leaner searchable tier for daily work.
Detection coverage mapping
The attack-technique knowledge base most teams use added data-component mappings, reaching version 18 in October 2025 with 106 Enterprise components and 120 in total. That helps map log sources to techniques, but the Atlas warns of a quiet failure mode, where a rule authored against one type of log source stops working when real data differs from what it expects.
What the numbers say about cost
Cost remains the loudest driver of change. The Atlas works through an example in which a large firewall feed held entirely in premium storage runs near 400 units daily, while a split design lands near 164 and extends raw history more than fourfold. It is only an illustration, but it explains why finance teams have taken an interest. Official advice backs the instinct to be selective, since Australian-led guidance observes that gathering too much leaves rule writers with a harder task.
One caution is worth repeating. Shortening retention below a service’s included window saves nothing; the Atlas notes the first 31 days are included in one major monitoring service. Always read the price sheet before announcing a saving.
Lessons the sport aviation community will recognise
Light sport owners know that a new standard is only as good as the logbook that proves compliance. Two cases in the Atlas make the same point for digital systems. One 2023 episode involved a government mail service, where the intrusion surfaced because an access record that lived only on a paid licence tier happened to be switched on and a bespoke rule watched it. A federal review panel afterwards urged that these records come at no extra charge and be kept for no fewer than six months, and the provider extended its standard audit tier to cover email access and a 180-day default. In 2024, a run of data thefts from a cloud warehouse involved stolen credentials on accounts with no multi-factor sign-in, and the activity left a recognisable query pattern: a temporary stage, a copy into it and a download. Both cases hinge on whether the right logs existed at the right time.
For a practical parallel, consider the routine discussed in the owner-focused rundown of landing gear inspection points, where the record of what was inspected matters as much as the inspection. A similar message runs through our notes on avoidable slips with LSA fuel systems, which stress catching small faults before they compound.
A maturity ladder for judging your own position
The Atlas proposes five stages: Reactive, Inventoried, Contracted, Measured and Optimising. A Reactive team finds out about a missing log during an incident. An Inventoried team has a source list and owners. A Contracted team has written agreements per source, including a freshness target and a test event. A Measured team monitors health and rehearses failure, for example with a monthly silence test. An Optimising team tunes tiers and costs with replay checks. Most organisations are somewhere between the first two, which is not a criticism so much as a baseline.
What to do in the next ninety days
The Atlas gives a rhythm that suits a modest team. In the first thirty days, build a register, score each source, set a retention policy and put monitors on the top ten sources. In the next thirty, write contracts, add fixtures and tests, and attach a price to each stream. In the final thirty, re-tier the three largest streams after a replay, perform an initial silence test and publish a short baseline.
Outlook
Expect more consolidation. Open schemas will keep absorbing vendor models, cheaper tiers will keep appearing, and regulators will keep asking how long evidence is retained. The label may change, and the Atlas is candid that it might. The underlying task, keeping complete, trustworthy and affordable evidence, is not going away. For light sport operators and anyone else who depends on records to prove a thing was done, that is a familiar and worthwhile goal.

