Billing axes and retention
Wakora bills on a few axes that map to things you can point at, not on metrics, sensors or data points. This page explains what each axis counts and how; the numbers per plan are on the pricing page and are not repeated here.
The axes
Hosts. Every machine with a registered agent that is not removed. A host that is switched off still counts until you remove it; a removed host does not. The first hosts are free on every workspace.
Monitored sites. One per distinct website the agent discovered and watches: an nginx
or Apache virtual host, an IIS site, folded to its bare domain (www.example.com and
example.com are one site), plus external sites you enrolled for RUM with the hosted
snippet and synthetic transaction flows. Aliases of one site and catch-all entries do not
count. A domain that only redirects elsewhere does count, because its certificate, its
DNS and its registration are watched like any other site's (the usage tab shows these
redirect-only domains separately); a bare domain bouncing to its own www is one site,
not a redirect. A dense box with a thousand sites bills on its sites, not as one host.
APM sites. A site becomes an APM site only when you enable it explicitly. Traces flow host-wide in the background; the switch decides which sites get a card, alerts and a billable slot. Switching a site off stops its billing at once and keeps its history.
Devices. One per approved network device polled over SNMP: a router, a switch, a UPS. A 48-port switch is one device; its ports, sensors and PoE readings are metrics within it, never counted separately. The collector host counts as a host.
Container groups. Containers are grouped by image, and a group counts only while at least one of its containers is running. Ten replicas of one image are one unit; a stopped group is free. Virtual machines and LXC guests are not a unit at all: a guest you monitor runs its own agent and is a host.
RUM views. Page views received from enrolled sites, cumulative over the calendar month.
Log volume. Gigabytes of log lines shipped per day, averaged over the month. Only lines that pass the agent's level filter count, so the default of notice and above keeps most workspaces far below their allowance; enabling full depth for a service is what moves this axis.
Flow volume. Gigabytes of network flow records (NetFlow, IPFIX) received per day, averaged over the month, for workspaces that use the collector's flow listener.
Status pages. The number of public status pages.
Answered phone calls. Alert calls a person picked up, per billing cycle. Unanswered, busy and failed calls are free, verification calls are never counted, and each plan with phone alerting includes a number of answered calls; beyond them a call budget you set yourself applies (see Billing, Phone calls).
How usage is measured
The console's Billing, Usage tab shows each axis live, and once a day a snapshot is taken. A monthly amount for a counted axis is the average of the daily snapshots, not the peak, so a host that existed for a day does not bill for the month. RUM views are the exception: they accumulate over the month, so the month's figure is the final count.
Going over a plan's allowance
An allowance is a wall for a workspace without a card on file: adding the next host or enabling the next APM site tells you the plan is full and points at the plans. With a card on file, the same action goes through and the unit is billed per month at the plan's unit price, shown on the button before you click it. Overage is linear and deliberately has no cap; the usage tab is where you keep an eye on it.
Agency workspaces bill wholesale: the agency pays for the usage of all its client workspaces at the wholesale unit prices, the clients see no billing at all. See Agencies.
Retention
Every plan keeps data for a fixed window, and the window differs by kind:
- Metrics and events are visible for the plan's retention period. Longer history is stored beyond it, so an upgrade backfills instantly instead of starting from the upgrade day. A downgrade narrows the window.
- Log search depth has its own, shorter window per plan, because logs are the heaviest data.
- Traces and profiles follow the metrics window.
- Heartbeats, which drive availability history, are kept long enough for the year strips on every plan.
Historical reads through the console and the API are clamped to the window; a request beyond it returns what the plan keeps. Where a year strip shows locked days, that is the retention boundary, not missing data.
What is never counted
Metrics per host, custom metrics, alert rules, notifications, incidents, team members, API calls, definitions from the marketplace. None of these carry a price; the plan decides which features exist, not how often you may use them.