Frequently asked questions

Metrics

What are Active Series?

Your Metrics usage is currently calculated based on one component: active series. An active series is a single time series that has received new data in the last 15 to 30 minutes. A time series is every unique combination of a Prometheus metric and all its labels.

Example

cpu_seconds_total{instance_type='small', region='west'}
cpu_seconds_total{instance_type='small', region='east'}

Here, the combination of the metric 'cpu_seconds_total' with different region labels ('west', 'east') has created two distinct active series, each representing the CPU usage of 'small' instances in different regions. As long as the series keep receiving new data, they keep counting as active.

Billing

How is my Metrics usage calculated?

We measure your active series during your billing cycle, and calculate the 95th percentile over the entire cycle. If your contract has included active series, this is subtracted from the amount that was measured.

Example

You have 500 metrics, each with 20 unique labels, all have received data over the last 15 minutes. This calculates to 500 * 20 = 10.000 active series.

Your contract includes 2.000 active series. This calculates to 10.000 - 2.000 = 8.000 billable active series.

Active series are billed per 1000 active series. If the current price is 5 EUR per 1000 active series, the billed amount comes to 8.000 / 1000 * 5 = 40 EUR.

How does a 95th percentile work for Metrics?

In our last example we assumed that your metrics usage was constant the entire month. In reality, your usage will fluctuate. For instance, you may have changed the amount of metrics gathered, exporters may have been added or removed, or there simply has been a spike in metrics because of high traffic on your systems or because of a misconfiguration.

To account for these fluctuations, we calculate the 95 percentile over the entire month. This means that all spikes that are in the top 5% of that month are ignored, and do not count towards your billing. Our billing cycle is one month. Assuming there are 720 hours in a month, this means we will ignore the top 36 hours (5% of 720) of that month.

Example

Your average amount of active series over the month is 5.000. At one point in the month you added a new exporter that produced a huge amount of metrics, and you spiked to 50.000 metrics over the course of 24 hours. You realized this problem the next day, and quickly disabled the exporter for the rest of the month.

Because your spike was shorter than 36 hours, you will be billed for 5.000 metrics that billing period.

How is my Logs usage calculated?

Your Logs usage is currently calculated based on one component: GiB stored. All logs that are ingested by our platform are processed and stored, and the actual storage consumed is billed at the end of the billing period.

We measure your storage usage during your billing cycle, and calculate the 95th percentile over the entire cycle. Even though logs normally don't experience the same kind of spikes as metrics, we still want to insulate you from common mistakes by ignoring the top 5%.

Where can I see my current usage?

Login to MyCyso using the same credentials that you use for Observe, and look for the Usage dashboard.

How does the billing cycle work?

Every billing cycle starts at the first day of the month, and ends on the last. During this period, your usage data is collected twice a day: during the morning and early evening. Exact times may vary a bit, but should be around 07:00 and 17:00 (UTC).

Every third day of the month, your usage is collected into a report, and an invoice is created. If you've opted for SEPA billing, the invoice will be automatically withdrawn from your account within a week.