What if the log management bill you budgeted for turns out to be half of what you really owe? It happens more often than most teams admit. The sticker price per gigabyte looks reasonable, the free tier feels generous, and then the invoice arrives with charges for queries, retention, indexing, and egress that nobody modeled. Cloud log management pricing is one of the few software categories where the headline number is almost never the real number.
This comparison breaks down how the major platforms actually charge, where the hidden costs hide, and how to build a cost model that survives contact with production traffic.
Why Per-GB Pricing Is Never Just Per-GB
Every vendor advertises a rate per gigabyte ingested. That figure is real, but it describes only one of at least four cost dimensions: ingestion, storage, query or scanning, and retention. A platform that charges $0.10 per GB to ingest can still cost more than one charging $2.50 per GB if the first platform bills heavily for indexing and long retention.
The practical mistake teams make is comparing ingestion rates in isolation. A better habit is to build a simple spreadsheet with four rows (ingest, store, query, retain) and multiply each by your real monthly volume. Most organizations discover that 60 to 80 percent of the bill comes from one or two dimensions, and it is rarely ingestion alone.
There is also a regional problem. Cloud providers price by region, and the same service can differ meaningfully between, say, US East and a European region. Any comparison that does not state the region and the date of price capture is close to useless six months later.
How the Major Platforms Charge

AWS CloudWatch Logs
CloudWatch splits billing into three separate meters. Standard ingestion ran about $0.50 per GB in published 2024 pricing. Archival storage sat around $0.03 per GB-month. Logs Insights queries cost roughly $0.005 per GB scanned. That last number deserves attention: a team that runs frequent ad-hoc queries over large volumes can spend more on scanning than on storing the data. The lesson is to scope queries tightly and avoid full-volume scans as a habit.
Azure Monitor Logs
Azure uses a pay-as-you-go per-GB ingestion model, roughly $2.30 to $2.76 per GB depending on region and tier in 2024, plus commitment tiers that lower the per-GB rate in exchange for a daily minimum commitment. Commitment tiers are excellent if your volume is stable and painful if it fluctuates, because you pay for the commitment whether you use it or not. Teams with spiky workloads should model both paths before signing.
Google Cloud Logging
Google historically charged about $0.50 per GB for ingestion above a free allotment, with the first 50 GB per month free, and about $0.01 per GB-month for storage beyond 30 days. Google has since revised this model, so treat those figures as historical rather than current. The generous free tier is genuinely useful for small teams, but it also creates a trap: the moment you cross the threshold, the bill steps up rather than ramps.
Splunk
Splunk is priced primarily by daily data volume ingested under a license model, not per-GB metered cloud billing. This makes direct per-GB comparison misleading. A Splunk license for a fixed daily volume behaves like a subscription with a ceiling, which rewards teams that can control what they send. Comparing Splunk’s effective per-GB cost to a metered cloud service is possible but requires normalizing for the committed volume and the fact that you pay for the ceiling, not the usage.
Datadog Log Management
Datadog charges per GB ingested (about $0.10 per GB in 2024) plus per-million-event charges for indexing and for 15-day retention, with longer retention billed separately. The low ingest rate is real, but the indexing and retention meters are where budgets move. Teams that index everything by default often find the per-million-event line item dominates.
Open-Source Stacks
Grafana Loki, OpenSearch, and the Elastic Stack carry no per-GB license fee. The cost shifts entirely to compute, storage, and engineering labor. A self-hosted stack can be dramatically cheaper at high volume and dramatically more expensive at low volume once you account for the salary hours spent operating it. The honest comparison is total cost of ownership, and that depends heavily on whether you self-host or buy a managed service.
Side-by-Side Comparison
| Platform | Ingest model | Approx. ingest rate (2024) | Storage / retention | Query cost | Main cost driver |
|---|---|---|---|---|---|
| AWS CloudWatch Logs | Per GB metered | ~$0.50/GB | ~$0.03/GB-month archival | ~$0.005/GB scanned | Query volume and scanning |
| Azure Monitor Logs | Per GB + commitment tiers | ~$2.30–$2.76/GB | Included per tier | Included in tier | Daily commitment level |
| Google Cloud Logging | Per GB above free tier | ~$0.50/GB (historical) | ~$0.01/GB-month past 30 days | Varies by model | Volume past free allotment |
| Splunk | Daily volume license | License-based | License-based | License-based | Committed daily volume |
| Datadog Logs | Per GB + per-million events | ~$0.10/GB | Per-million-event retention fees | Indexing fees | Indexing and retention |
| Open source (Loki, OpenSearch) | Self-hosted | No license fee | Your infrastructure cost | Your compute cost | Engineering labor and compute |
The table makes one thing obvious: no two vendors meter the same way. That is why a straight per-GB comparison across all six rows would be dishonest.
The Hidden Costs Nobody Models
Egress is the classic surprise. Moving logs out of a cloud region or into a third-party tool can trigger transfer charges that never appear in the ingestion rate. Retrieval from cold storage is the second classic. Archival storage like S3 Glacier or Azure Archive is far cheaper per GB-month than hot storage, but retrieving and querying that data can cost more than the storage saved. Teams that archive aggressively without modeling retrieval get burned the first time an auditor asks for a six-month-old log.
Indexing is the third. Any platform that charges per indexed event or per million events rewards selective indexing. The fix is straightforward: index what you query, store the rest in a cheaper tier, and route by log type rather than indexing everything by default.
Free tiers are the fourth. Google’s 50 GB per month allotment, Azure’s included ingestion per commitment tier, and AWS’s limited free tier are not apples-to-apples. A free tier that covers your whole workload today may cover a third of it after a product launch.
A Practical Method for Comparing Costs
- Measure real monthly volume in GB for at least 30 days, split by log source.
- Estimate query load separately: how many GB get scanned or indexed per month, and how often.
- Define retention requirements by log type, not as a single blanket number.
- Price each vendor on all four meters using the same volume and query assumptions.
- Add a 20 percent buffer for growth and for the meters you inevitably underestimated.
This method takes an afternoon and routinely saves five figures a year. It also makes vendor negotiations concrete, because you can point to the specific meter that is driving cost.
What Compliance Does to the Bill
Retention is often not a business choice at all. Compliance frameworks dictate minimums. PCI DSS, for example, requires 12 months of audit log retention with 3 months immediately available. That single requirement can multiply storage costs, because “immediately available” means hot or warm storage, not archive. Before comparing vendors, map every regulatory obligation to a retention tier. A platform that charges per indexed event for long retention will price a PCI workload very differently from one that charges flat storage.
Cost governance is its own discipline. FinOps teams routinely find that 20 to 30 percent of log volume is debug-level noise that nobody queries. Dropping or sampling that traffic is often the single largest cost reduction available, and it works on every platform in this comparison.
If you are comparing adjacent tool categories, the same metering logic applies. The way vendors structure per-seat versus per-usage fees in social media management tools pricing comparison for teams mirrors the ingestion-versus-indexing split in log platforms. The same is true when you look at how campaign management tools pricing comparison you actually pay for usage tiers, or how affiliate marketing tools pricing comparison flat fees vs usage-based models play out over a year. The pattern repeats: the headline rate is never the whole bill.
Common Mistakes and What Works in Practice
The most common error is comparing list prices without normalizing volume. The second is ignoring query patterns, which is where CloudWatch and Datadog bills most often surprise teams. The third is treating the free tier as permanent. The fourth is archiving without a retrieval plan.
What actually works: tag every log stream with an owner and a retention class, set ingestion filters at the agent level rather than the platform level, and review the top ten log sources by volume every quarter. Teams that do this consistently cut 15 to 25 percent from their bill without losing anything they query.
A related trap appears in adjacent categories too. The way proposal tools pricing comparison you actually pay structures seat minimums is the same pattern as Azure commitment tiers. The same logic shows up in content audit tools pricing comparison you really pay and in AI voice cloning tools pricing comparison you actually pay, where per-minute or per-character meters hide behind low entry prices. Even AI diagram tools pricing comparison you actually pay per and AI translation tools pricing comparison you actually pay follow the same rule: model the meter that scales with usage, not the one that looks cheapest on the pricing page.
Conclusion
Cloud log management pricing rewards teams that model all four meters instead of fixating on the ingest rate. AWS charges for scanning, Azure charges for commitment, Google charges past a free allotment, Splunk charges for a daily ceiling, Datadog charges for indexing and retention, and open source charges in engineering hours. None of those models is inherently cheaper; each is cheaper for a particular workload shape.
The winning approach is unglamorous: measure real volume, split by log source, map compliance retention to storage tiers, price every vendor on the same assumptions, and add a buffer. Do that, and the invoice stops being a surprise. Skip it, and the cheapest-looking platform on the pricing page is often the most expensive one on the bill.
FAQ
Which cloud log management tool is cheapest per GB?
On raw ingestion, Datadog’s roughly $0.10 per GB looks lowest, followed by AWS and Google at around $0.50 per GB. But ingestion is only one meter. Datadog adds per-million-event indexing and retention fees, AWS adds query scanning fees, and Google charges only above its free allotment. The cheapest platform depends on your query volume and retention needs, not the ingest rate alone.
Is Splunk more expensive than cloud-native log tools?
It is not directly comparable, because Splunk is licensed by daily ingest volume rather than metered per GB. A Splunk license behaves like a subscription with a ceiling, so the effective cost depends on how close you run to that ceiling. Teams with stable, predictable volume often find it competitive. Teams with spiky volume often find metered cloud pricing cheaper.
How much does log retention actually cost?
It depends on the tier. Hot storage in a log platform is expensive per GB-month, while archival storage is far cheaper but carries retrieval and query costs. The practical rule is to keep the compliance window in hot or warm storage and push everything older to archive, then model retrieval costs before assuming archive is cheaper overall.
Are free tiers enough for a small team?
Sometimes, but treat them as temporary. Google’s 50 GB per month logging allotment covers many small teams entirely. Azure’s included ingestion per commitment tier and AWS’s limited free tier are narrower. The risk is that a free tier covers your workload until a product launch doubles your volume, at which point the bill steps up sharply rather than gradually.
What is the biggest hidden cost in log management?
Query and scanning costs surprise teams most often. AWS charges about $0.005 per GB scanned by Logs Insights, and frequent ad-hoc queries over large volumes can exceed storage costs. Indexing fees on per-event platforms run a close second. Scoping queries tightly and indexing selectively are the two biggest fixes.
How does compliance affect log management pricing?
Frameworks like PCI DSS require 12 months of audit log retention with 3 months immediately available. That “immediately available” clause forces a portion of your logs into hot or warm storage, which is the expensive tier. Mapping each regulatory requirement to a storage tier before comparing vendors prevents underestimating the bill by a wide margin.