2023-01-04 07:19:18 +01:00
|
|
|
package streamaggr
|
|
|
|
|
|
|
|
import (
|
|
|
|
"fmt"
|
|
|
|
"strings"
|
|
|
|
"testing"
|
|
|
|
|
|
|
|
"github.com/VictoriaMetrics/VictoriaMetrics/lib/prompbmarshal"
|
|
|
|
)
|
|
|
|
|
|
|
|
func BenchmarkAggregatorsPushByJobAvg(b *testing.B) {
|
|
|
|
for _, output := range []string{
|
|
|
|
"total",
|
lib/streamaggr: huge pile of changes
- Reduce memory usage by up to 5x when de-duplicating samples across big number of time series.
- Reduce memory usage by up to 5x when aggregating across big number of output time series.
- Add lib/promutils.LabelsCompressor, which is going to be used by other VictoriaMetrics components
for reducing memory usage for marshaled []prompbmarshal.Label.
- Add `dedup_interval` option at aggregation config, which allows setting individual
deduplication intervals per each aggregation.
- Add `keep_metric_names` option at aggregation config, which allows keeping the original
metric names in the output samples.
- Add `unique_samples` output, which counts the number of unique sample values.
- Add `increase_prometheus` and `total_prometheus` outputs, which ignore the first sample
per each newly encountered time series.
- Use 64-bit hashes instead of marshaled labels as map keys when calculating `count_series` output.
This makes obsolete https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5579
- Expose various metrics, which may help debugging stream aggregation:
- vm_streamaggr_dedup_state_size_bytes - the size of data structures responsible for deduplication
- vm_streamaggr_dedup_state_items_count - the number of items in the deduplication data structures
- vm_streamaggr_labels_compressor_size_bytes - the size of labels compressor data structures
- vm_streamaggr_labels_compressor_items_count - the number of entries in the labels compressor
- vm_streamaggr_flush_duration_seconds - a histogram, which shows the duration of stream aggregation flushes
- vm_streamaggr_dedup_flush_duration_seconds - a histogram, which shows the duration of deduplication flushes
- vm_streamaggr_flush_timeouts_total - counter for timed out stream aggregation flushes,
which took longer than the configured interval
- vm_streamaggr_dedup_flush_timeouts_total - counter for timed out deduplication flushes,
which took longer than the configured dedup_interval
- Actualize docs/stream-aggregation.md
The memory usage reduction increases CPU usage during stream aggregation by up to 30%.
This commit is based on https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5850
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5898
2024-03-02 01:42:26 +01:00
|
|
|
"total_prometheus",
|
2023-01-04 07:19:18 +01:00
|
|
|
"increase",
|
lib/streamaggr: huge pile of changes
- Reduce memory usage by up to 5x when de-duplicating samples across big number of time series.
- Reduce memory usage by up to 5x when aggregating across big number of output time series.
- Add lib/promutils.LabelsCompressor, which is going to be used by other VictoriaMetrics components
for reducing memory usage for marshaled []prompbmarshal.Label.
- Add `dedup_interval` option at aggregation config, which allows setting individual
deduplication intervals per each aggregation.
- Add `keep_metric_names` option at aggregation config, which allows keeping the original
metric names in the output samples.
- Add `unique_samples` output, which counts the number of unique sample values.
- Add `increase_prometheus` and `total_prometheus` outputs, which ignore the first sample
per each newly encountered time series.
- Use 64-bit hashes instead of marshaled labels as map keys when calculating `count_series` output.
This makes obsolete https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5579
- Expose various metrics, which may help debugging stream aggregation:
- vm_streamaggr_dedup_state_size_bytes - the size of data structures responsible for deduplication
- vm_streamaggr_dedup_state_items_count - the number of items in the deduplication data structures
- vm_streamaggr_labels_compressor_size_bytes - the size of labels compressor data structures
- vm_streamaggr_labels_compressor_items_count - the number of entries in the labels compressor
- vm_streamaggr_flush_duration_seconds - a histogram, which shows the duration of stream aggregation flushes
- vm_streamaggr_dedup_flush_duration_seconds - a histogram, which shows the duration of deduplication flushes
- vm_streamaggr_flush_timeouts_total - counter for timed out stream aggregation flushes,
which took longer than the configured interval
- vm_streamaggr_dedup_flush_timeouts_total - counter for timed out deduplication flushes,
which took longer than the configured dedup_interval
- Actualize docs/stream-aggregation.md
The memory usage reduction increases CPU usage during stream aggregation by up to 30%.
This commit is based on https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5850
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5898
2024-03-02 01:42:26 +01:00
|
|
|
"increase_prometheus",
|
2023-01-04 07:19:18 +01:00
|
|
|
"count_series",
|
|
|
|
"count_samples",
|
lib/streamaggr: huge pile of changes
- Reduce memory usage by up to 5x when de-duplicating samples across big number of time series.
- Reduce memory usage by up to 5x when aggregating across big number of output time series.
- Add lib/promutils.LabelsCompressor, which is going to be used by other VictoriaMetrics components
for reducing memory usage for marshaled []prompbmarshal.Label.
- Add `dedup_interval` option at aggregation config, which allows setting individual
deduplication intervals per each aggregation.
- Add `keep_metric_names` option at aggregation config, which allows keeping the original
metric names in the output samples.
- Add `unique_samples` output, which counts the number of unique sample values.
- Add `increase_prometheus` and `total_prometheus` outputs, which ignore the first sample
per each newly encountered time series.
- Use 64-bit hashes instead of marshaled labels as map keys when calculating `count_series` output.
This makes obsolete https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5579
- Expose various metrics, which may help debugging stream aggregation:
- vm_streamaggr_dedup_state_size_bytes - the size of data structures responsible for deduplication
- vm_streamaggr_dedup_state_items_count - the number of items in the deduplication data structures
- vm_streamaggr_labels_compressor_size_bytes - the size of labels compressor data structures
- vm_streamaggr_labels_compressor_items_count - the number of entries in the labels compressor
- vm_streamaggr_flush_duration_seconds - a histogram, which shows the duration of stream aggregation flushes
- vm_streamaggr_dedup_flush_duration_seconds - a histogram, which shows the duration of deduplication flushes
- vm_streamaggr_flush_timeouts_total - counter for timed out stream aggregation flushes,
which took longer than the configured interval
- vm_streamaggr_dedup_flush_timeouts_total - counter for timed out deduplication flushes,
which took longer than the configured dedup_interval
- Actualize docs/stream-aggregation.md
The memory usage reduction increases CPU usage during stream aggregation by up to 30%.
This commit is based on https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5850
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5898
2024-03-02 01:42:26 +01:00
|
|
|
"unique_samples",
|
2023-01-04 07:19:18 +01:00
|
|
|
"sum_samples",
|
|
|
|
"last",
|
|
|
|
"min",
|
|
|
|
"max",
|
|
|
|
"avg",
|
|
|
|
"stddev",
|
|
|
|
"stdvar",
|
|
|
|
"histogram_bucket",
|
|
|
|
"quantiles(0, 0.5, 1)",
|
|
|
|
} {
|
|
|
|
b.Run(fmt.Sprintf("output=%s", output), func(b *testing.B) {
|
|
|
|
benchmarkAggregatorsPush(b, output)
|
|
|
|
})
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
func benchmarkAggregatorsPush(b *testing.B, output string) {
|
|
|
|
config := fmt.Sprintf(`
|
|
|
|
- match: http_requests_total
|
|
|
|
interval: 24h
|
|
|
|
without: [job]
|
|
|
|
outputs: [%q]
|
|
|
|
`, output)
|
2024-02-29 01:39:00 +01:00
|
|
|
pushFunc := func(tss []prompbmarshal.TimeSeries) {}
|
2024-01-24 11:31:27 +01:00
|
|
|
a, err := newAggregatorsFromData([]byte(config), pushFunc, 0)
|
2023-01-04 07:19:18 +01:00
|
|
|
if err != nil {
|
|
|
|
b.Fatalf("unexpected error when initializing aggregators: %s", err)
|
|
|
|
}
|
|
|
|
defer a.MustStop()
|
|
|
|
|
lib/streamaggr: huge pile of changes
- Reduce memory usage by up to 5x when de-duplicating samples across big number of time series.
- Reduce memory usage by up to 5x when aggregating across big number of output time series.
- Add lib/promutils.LabelsCompressor, which is going to be used by other VictoriaMetrics components
for reducing memory usage for marshaled []prompbmarshal.Label.
- Add `dedup_interval` option at aggregation config, which allows setting individual
deduplication intervals per each aggregation.
- Add `keep_metric_names` option at aggregation config, which allows keeping the original
metric names in the output samples.
- Add `unique_samples` output, which counts the number of unique sample values.
- Add `increase_prometheus` and `total_prometheus` outputs, which ignore the first sample
per each newly encountered time series.
- Use 64-bit hashes instead of marshaled labels as map keys when calculating `count_series` output.
This makes obsolete https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5579
- Expose various metrics, which may help debugging stream aggregation:
- vm_streamaggr_dedup_state_size_bytes - the size of data structures responsible for deduplication
- vm_streamaggr_dedup_state_items_count - the number of items in the deduplication data structures
- vm_streamaggr_labels_compressor_size_bytes - the size of labels compressor data structures
- vm_streamaggr_labels_compressor_items_count - the number of entries in the labels compressor
- vm_streamaggr_flush_duration_seconds - a histogram, which shows the duration of stream aggregation flushes
- vm_streamaggr_dedup_flush_duration_seconds - a histogram, which shows the duration of deduplication flushes
- vm_streamaggr_flush_timeouts_total - counter for timed out stream aggregation flushes,
which took longer than the configured interval
- vm_streamaggr_dedup_flush_timeouts_total - counter for timed out deduplication flushes,
which took longer than the configured dedup_interval
- Actualize docs/stream-aggregation.md
The memory usage reduction increases CPU usage during stream aggregation by up to 30%.
This commit is based on https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5850
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5898
2024-03-02 01:42:26 +01:00
|
|
|
const loops = 10
|
2024-02-29 01:39:00 +01:00
|
|
|
|
2023-01-04 07:19:18 +01:00
|
|
|
b.ReportAllocs()
|
2024-02-29 01:39:00 +01:00
|
|
|
b.SetBytes(int64(len(benchSeries) * loops))
|
2023-01-04 07:19:18 +01:00
|
|
|
b.RunParallel(func(pb *testing.PB) {
|
2023-07-25 01:44:09 +02:00
|
|
|
var matchIdxs []byte
|
2023-01-04 07:19:18 +01:00
|
|
|
for pb.Next() {
|
2024-02-29 01:39:00 +01:00
|
|
|
for i := 0; i < loops; i++ {
|
lib/streamaggr: huge pile of changes
- Reduce memory usage by up to 5x when de-duplicating samples across big number of time series.
- Reduce memory usage by up to 5x when aggregating across big number of output time series.
- Add lib/promutils.LabelsCompressor, which is going to be used by other VictoriaMetrics components
for reducing memory usage for marshaled []prompbmarshal.Label.
- Add `dedup_interval` option at aggregation config, which allows setting individual
deduplication intervals per each aggregation.
- Add `keep_metric_names` option at aggregation config, which allows keeping the original
metric names in the output samples.
- Add `unique_samples` output, which counts the number of unique sample values.
- Add `increase_prometheus` and `total_prometheus` outputs, which ignore the first sample
per each newly encountered time series.
- Use 64-bit hashes instead of marshaled labels as map keys when calculating `count_series` output.
This makes obsolete https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5579
- Expose various metrics, which may help debugging stream aggregation:
- vm_streamaggr_dedup_state_size_bytes - the size of data structures responsible for deduplication
- vm_streamaggr_dedup_state_items_count - the number of items in the deduplication data structures
- vm_streamaggr_labels_compressor_size_bytes - the size of labels compressor data structures
- vm_streamaggr_labels_compressor_items_count - the number of entries in the labels compressor
- vm_streamaggr_flush_duration_seconds - a histogram, which shows the duration of stream aggregation flushes
- vm_streamaggr_dedup_flush_duration_seconds - a histogram, which shows the duration of deduplication flushes
- vm_streamaggr_flush_timeouts_total - counter for timed out stream aggregation flushes,
which took longer than the configured interval
- vm_streamaggr_dedup_flush_timeouts_total - counter for timed out deduplication flushes,
which took longer than the configured dedup_interval
- Actualize docs/stream-aggregation.md
The memory usage reduction increases CPU usage during stream aggregation by up to 30%.
This commit is based on https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5850
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5898
2024-03-02 01:42:26 +01:00
|
|
|
series := benchSeries
|
|
|
|
for len(series) > 0 {
|
|
|
|
chunk := series
|
|
|
|
if len(chunk) > 1_000 {
|
|
|
|
chunk = series[:1_000]
|
|
|
|
series = series[len(chunk):]
|
|
|
|
} else {
|
|
|
|
series = nil
|
|
|
|
}
|
|
|
|
matchIdxs = a.Push(chunk, matchIdxs)
|
|
|
|
}
|
2024-02-29 01:39:00 +01:00
|
|
|
}
|
2023-01-04 07:19:18 +01:00
|
|
|
}
|
|
|
|
})
|
|
|
|
}
|
|
|
|
|
2024-02-29 01:39:00 +01:00
|
|
|
func newBenchSeries(seriesCount int) []prompbmarshal.TimeSeries {
|
|
|
|
a := make([]string, seriesCount)
|
|
|
|
for j := 0; j < seriesCount; j++ {
|
|
|
|
s := fmt.Sprintf(`http_requests_total{path="/foo/%d",job="foo",instance="bar",pod="pod-123232312",namespace="kube-foo-bar",node="node-123-3434-443",`+
|
|
|
|
`some_other_label="foo-bar-baz",environment="prod",label1="value1",label2="value2",label3="value3"} %d`, j, j*1000)
|
|
|
|
a = append(a, s)
|
2023-01-04 07:19:18 +01:00
|
|
|
}
|
|
|
|
metrics := strings.Join(a, "\n")
|
|
|
|
return mustParsePromMetrics(metrics)
|
|
|
|
}
|
|
|
|
|
2024-02-29 01:39:00 +01:00
|
|
|
const seriesCount = 10_000
|
2023-01-04 07:19:18 +01:00
|
|
|
|
2024-02-29 01:39:00 +01:00
|
|
|
var benchSeries = newBenchSeries(seriesCount)
|