Model ingestion volume from hosts, apps, and devices. Tune intervals, tags, compression, and retention choices. See costs instantly, export reports, and optimize spend safely.
| Scenario | Hosts | Apps | Devices | Interval (sec) | Tags | Retention (days) | Estimated samples/month |
|---|---|---|---|---|---|---|---|
| Small | 10 | 5 | 20 | 60 | 6 | 7 | ≈ 6,307,200 |
| Medium | 50 | 20 | 100 | 15 | 8 | 15 | ≈ 344,064,000 |
| Large | 250 | 120 | 2,000 | 10 | 10 | 30 | ≈ 7,776,000,000 |
Metrics ingestion spend is driven by three levers: how many active series you emit, how often you sample them, and how long you retain them. Series count is the multiplier, because labels and high-cardinality dimensions can turn one metric name into thousands of distinct streams. This calculator converts those choices into samples per day and samples per month, then applies your per‑million ingestion rate and any regional multiplier to produce a budgetable estimate.
Sampling interval sets the velocity of points. Moving from 60 seconds to 15 seconds increases samples roughly four times, even if series count stays constant. Drop or filter rate represents relabeling, downsampling, or selective forwarding at the agent or gateway. Use it to model reductions from noisy metrics, redundant dimensions, or environments you exclude. When you compare scenarios, keep the interval consistent first, then adjust drops to test governance policies.
Not all samples cost the same to store. Tags increase payload size and indexing work, which is why the calculator asks for tags per metric, average tag bytes, a base sample size, and an index overhead percentage. Compression reduces bytes, but tag sets and metadata can offset the gains. If your platform bills on ingestion, these fields matter for storage estimates; if it bills on bytes ingested, increase base and tag bytes to mirror the encoder.
Retention determines steady‑state storage because every day of data adds another day’s worth of samples until the window is full. The calculator approximates stored samples as daily samples multiplied by retention days, then converts effective bytes to GB‑months. Use this section to align monitoring needs with tiers: shorter retention for high‑resolution troubleshooting, longer retention for SLO reporting and trend analysis. A mixed strategy often lowers cost without sacrificing insight.
Treat the output as a planning range, not a final invoice. Start with realistic series counts from your telemetry inventory, then model changes you can deploy: reduce label cardinality, standardize metric names, increase intervals for low‑value signals, and enforce retention by service class. Finally, test pricing levers such as commitments and region selection. Export CSV or PDF to share assumptions with finance, platform teams, and service owners.
A sample is one timestamped value for one unique time series. If a series is scraped every 15 seconds, it produces 5,760 samples per day. Multiply by the number of active series to estimate volume.
Labels create unique series combinations. Adding a high-cardinality label, like user_id, can multiply series counts dramatically, increasing ingestion, indexing, and storage. Prefer bounded labels such as region, status_code, or service.
Use representative tag keys and values from your telemetry. Count characters, include separators, then estimate overhead for indexing and metadata. If unsure, start with defaults, run a sensitivity test, and adjust until estimates align with observed storage usage.
Compression mainly affects stored footprint. Some platforms bill ingestion by samples, not bytes, so compression won’t change ingestion charges. If your provider bills by bytes ingested, raise base and tag bytes to match your encoded payload.
The calculator subtracts the free tier from monthly samples, then applies the discount factor to ingestion, API writes, and storage. Enter commitment discounts or negotiated rates here to compare scenarios consistently.
Export the report, then compare estimated series count and sample rates with your agent metrics and billing dashboard. Adjust drop rate, retention, and bytes-per-sample assumptions until the model matches a known month.
Important Note: All the Calculators listed in this site are for educational purpose only and we do not guarentee the accuracy of results. Please do consult with other sources as well.