Prometheus instant vector vs range vector

VictoriaMetrics author here. This is Prometheus-like monitoring system, which supports PromQL-like query language - MetricsQL.

The instant vector and range vector are indeed confusing terms in Prometheus. That's why these terms are avoided in VictoriaMetrics docs. Prometheus query language - PromQL - provides various functions, which can be divided into two groups:

• Functions, which accept only instant vector. Such functions can be split into the following subgroups:
  • transform functions, which apply various transformations individually per each input time series. For example, abs()
  • label manipulation functions, which modify labels and metric names for the input time series. For example, label_replace()
  • aggregate functions, which aggregate multiple input time series into specified groups of output time series. For example, sum(). Fun fact is that aggregate functions are named aggregation operators in Prometheus - see these docs.
• Functions, which accept only range vector. VictoriaMetrics names such functions as rollup functions, since they calculate the result based on input time series samples over the given lookbehind window specified in square brackets (aka sliding window). For example, rate(http_requests_total[5m]) calculates the average per-second increase rate for http_requests_total time series over the last 5 minutes.

From user's perspective the only difference between instant vector and range vector is that range vector is constructed from the instant vector by adding a lookbehind window in square brackets. For example, http_requests_total is an instant vector, while http_requests_total[5m] is a range vector. I'd say that the range vector syntax is just a syntactic sugar for rollup functions in PromQL. E.g. rate(m[d]) could be written as rate(m, d), e.g. the lookbehind window d could be passed as a separate argument to rollup functions.

You need to get familiar with two other related terms:

• Instant query: when you query Prometheus for result of an expression on a single timestamp. e.g. for alerting.
• Range query: when you query Prometheus for an expression with start and end timestamps. e.g. for graphing in Grafana.

So your expression can have a number of instant and range vectors in it, and be sent to Prometheus as an instant or range query.

Summary from

• Understanding Prometheus Range Vectors
• The Anatomy of a PromQL Query

• What’s a Vector?
  • Since Prometheus is a timeseries database, all data is in the context of some timestamp. The series that maps a timestamp to recorded data is called a timeseries
  • a set of related timeseries is called a vector
  • Ex.
    • http_requests_total is a vector representing the total number of http requests received by a service
    • http_requests_total{code="200"} http_requests_total refers to the entire set of timeseries that are named that. And by appending a {code="200"}, we’re selecting a subset.
• Types of Vectors
  • Instant vector - a set of timeseries where every timestamp maps to a single data point at that “instant”
    • Imagine evaluating the expression http_requests_total at a given timestamp. http_requests_total is an instant vector selector that selects the latest sample for any time series with the metric name http_requests_total. More specifically, "latest" means "at most 5 minutes old and not stale", relative to the evaluation timestamp. So this selector will only yield a result for series that have a sample at most 5 minutes prior to the evaluation timestamp, and where the last sample before the evaluation timestamp is not a stale marker (an explicit way of marking a series as terminating at a certain time in the Prometheus TSDB).
  • Range vector - a set of timeseries where every timestamp maps to a “range” of data points, recorded some duration into the past.
    • Range vector is mostly used for graphs, where you want to show a PromQL expression over a given time range. A range query works exactly like many completely independent instant queries that are evaluated at subsequent time steps over a given range of time. Of course, this is highly optimized under the hood and Prometheus doesn't actually run many independent instant queries in this case.
• Differences
  • Instant vectors can be charted; Range vectors cannot. This is because charting something involves displaying a data point on the y-axis for every timestamp on the x-axis. Instant vectors have a single value for every timestamp, while range vectors have many of them. For the purpose of charting a metric, it is undefined1 how to show multiple data points for a single timestamp in a timeseries.
  • Instant vectors can be compared and have arithmetic performed on them; Range vectors cannot. This is also due to the way comparison and arithmetic operators are defined. For every timestamp, if we have multiple values, we don’t know how to add1 or compare them to another timeseries of a similar nature.
  • Range Vectors for counters. We take the instant vector and append our duration [15m]. This part is called the range selector and it transforms the instant vector into a range vector. We then use a function like increase which effectively subtracts the data point at the start of the range from the one at the end. increase(http_requests_total{code="200",handler="/api/v1/query"}[15m]) represent it is the increase in the total number of requests over the past fifteen minutes

• I see the value of the metric with no timestamp : to answer this first, a time stamp is not shown in the results table, since we have the "Evaluation time" field already present Image for reference
• Range vector seems more logical as it has values per timestamp : range vectors works between a range of 2 timestamps, so when you apply the [5m] to your metric, it take now-5m of the range and displays the results in that time frame, and since there maybe more than 1 results in that time frame the timestamp is shown.

This video helped me understand this very same question, hope it helps you too.

https://training.robustperception.io/courses/204997/lectures/3156025