- Market
- Saudi Arabia
- Written for
- CISO, CTO
- Decision
- Design a logging and monitoring estate before committing to a platform
- Languages
- EN · AR
The bill
The four decisions that set what a logging estate costs
Log storage pricing looks like a platform question; it is actually the sum of four design decisions, each of which defaults expensively when nobody makes it.
| Variable | Effect on price | Why |
|---|---|---|
| Which sources you ingest | Sets the baseline volume everything else multiplies | The verbose network appliance connected in an afternoon can out-bill the identity provider that would actually witness an intrusion. Filtering a source before storage, or not connecting it at all, is a design decision — made per source, on what its records would prove. |
| How long you keep each source | Multiplies every unit of volume ingested | A single retention number applied to everything overpays for chatty sources and underinsures the quiet ones investigations depend on. The schedule is written per source, from how long intrusions go unnoticed and what an investigation would need to reconstruct. |
| Which tier each source lands in | Separates the cost of searching from the cost of keeping | Hot, searchable storage is priced for the queries analysts run daily; archive storage is priced for evidence read once, if ever. Keeping everything hot because tiering was never designed is the single most common overspend we find. |
| How the data is actually queried | Decides what must stay fast and what may sleep | The telemetry behind live triage has to answer in seconds; the audit trail a regulator may one day request does not. Watching real query patterns for a period, then tiering to match them, turns the bill into a reflection of use. |
Ask any supplier quoting this
- What does it cost, and how long does it take, to bring archived logs back when an investigation needs them?
- Which of your charges apply to data we filter or drop before it reaches storage?
The order of work
Log sources are ranked by attack path, not by ease of connection
The design starts from harm and works backwards. Where would a compromise hurt this organisation — the system that moves money, the store that holds customer data, the platform production runs on — and which systems would an attacker cross to reach it: the identity provider, the endpoints, the remote access path, the mail platform. Those crossings are where traces would exist, so those sources are onboarded first and parsed properly. Ranked this way, the list rarely matches the order of connection convenience; the appliance with the friendly export wizard drops down the list, and the awkward application audit trail nobody wants to parse moves up it.
Retention follows the same logic. The question is not what a default allows but what an investigation would need: intrusions are routinely discovered long after they begin, so the sources that would reconstruct an intruder's first steps need a longer memory than the ones that only describe yesterday. Regulated sectors in Saudi Arabia carry expectations about how long security records are kept, and those expectations are treated as a floor per source, not as the design; the schedule is reasoned above them where investigation need demands it, with the reason for each period written next to it so the next budget review argues with the logic rather than with a number nobody remembers choosing.
The shape
One telemetry stream, three storage decisions
Every source that enters the estate passes through the same policy, which decides — per source, on record — what stays searchable, what compresses, and what archives as evidence.
Four telemetry inputs — identity events, endpoint telemetry, network logs, application audit trails — converge on a tiering policy that produces fast investigation queries, a reopenable archive, and a governed ingest bill.
Identity events
Sign-ins, privilege changes, MFA outcomes.
Endpoint telemetry
Process activity, high volume, high value.
Network and perimeter logs
Verbose; filtered before they become a bill.
Application audit trails
The record of what happened inside the systems that matter.
What we do
The tiering policy
Per source: what is filtered at ingest, how long it stays hot for live queries, when it compresses to warm, when it archives — and what is deliberately not collected, named so the omission is a decision.
Fast queries for live investigation
The telemetry triage runs on answers in seconds.
An archive an incident can reopen
Evidence kept affordably, with a tested path back to searchable.
An ingest bill under governance
Spend that tracks visibility, reviewed as sources change.
The policy is a living document: a new system entering the estate gets a row before it gets a connector.
Found too late
The logging gaps that only an incident reveals
Each of these is invisible in a quiet estate and decisive in an investigation. Designing them out beforehand is most of what this engagement is for.
- 01
The source that was never onboarded
The system at the centre of the intrusion logged locally and rotated its files after a few days; by discovery, the record of the first access was gone. The attack-path ranking exists to catch this system before an attacker does.
- 02
The window that closes before the story starts
The intrusion began before the oldest log the estate kept, so the investigation could describe the middle of the attack but never its entry. Retention reasoned from dwell, per source, is the countermeasure.
- 03
The clocks that disagree
Sources stamped in different time zones, or drifting without synchronisation, make sequencing impossible — did the sign-in precede the file access or follow it. Time discipline is a design requirement, not an ops nicety.
- 04
The field the parser threw away
The events arrived, but ingestion dropped the one field the investigation needed — the client address behind the proxy, the original recipient of the forwarded mail. Verifying parsing per source, against an investigation scenario, closes this quietly fatal gap.
Our role
What we design in the logging estate and what your platform team runs
- 01
The source map with its reasons
A ranked onboarding plan traced from your attack paths, with parsing verified per source against an investigation scenario before the source counts as done.
- 02
A retention schedule you can defend
Periods reasoned per source from investigation need and the obligations that apply to you, each with its reason recorded — so the schedule survives both an audit and a budget review.
- 03
A tiering policy and a rehearsed way back
The per-source tier assignments, and a tested procedure for re-hydrating archived evidence — because an archive that has never been read back is a hope, and incidents are the wrong time to test hopes.
The platform choice, the budget envelope and the risk you will carry unlogged are your decisions; we put the design in front of you with its trade-offs priced in visibility, so those decisions are made once, in daylight, instead of nightly by default settings.
Common questions
What buyers ask
How long should security logs actually be kept?+
There is no honest single number, which is why we refuse to give one. The schedule is reasoned per source from three inputs: how long intrusions typically remain undiscovered — which argues for a long memory on the sources that would reconstruct an attacker's entry, such as identity and remote access logs; what an investigation would need from each source to rebuild a timeline; and the retention expectations of the regulators and contracts that apply to your sector, which set a floor rather than a design. A chatty network source may honestly need far less than a quiet audit trail. What matters is that each period has its reason written next to it, so the schedule can be defended and revised deliberately.
Is it wasteful to log systems nobody monitors in real time?+
No, because monitoring and evidence are different jobs with different economics. Real-time monitoring needs hot, searchable, expensive storage and a detection watching it; evidence needs to exist, cheaply, until the day an investigation asks what happened on that system in a given week. Plenty of sources earn archive-tier collection without earning a detection: the record costs little to keep cold and is irreplaceable when needed, because logging cannot be applied retroactively. The waste runs the other way — paying hot-tier prices for sources nobody queries, which is what happens when tiering is never designed.
Why do investigations stall even in estates that ingest heavily?+
Because volume is not visibility. The stalls we see repeat a pattern: the pivotal system was never onboarded because it was awkward to connect while easy sources filled the quota; the events arrived but parsing dropped the field the question turned on; clocks disagreed across sources so the sequence of events could not be established; or retention on the one source that held the entry point was shorter than the intruder's dwell. None of these are platform failures — every one is a design decision that was never made. A heavy ingest bill mostly proves that connecting sources is easier than choosing them.
In this cluster
Design the estate with us
A design session that drafts your source ranking, your retention schedule with its reasons, and your tiering policy — before any platform commitment is signed.
Start the conversation
