Add free function; Optimize read with aggregation

This commit is contained in:
Lou Knauer 2021-09-07 09:21:08 +02:00
parent cbe5a0714c
commit dc92d17675
2 changed files with 170 additions and 51 deletions

View File

@ -8,7 +8,7 @@ import (
// Default buffer capacity.
// `buffer.data` will only ever grow up to it's capacity and a new link
// in the buffer chain will be created if needed so that no copying
// of needs to happen on writes.
// of data or reallocation needs to happen on writes.
const (
BUFFER_CAP int = 1024
)
@ -20,6 +20,11 @@ var bufferPool sync.Pool = sync.Pool{
},
}
var (
ErrNoData error = errors.New("no data for this metric/level")
ErrDataDoesNotAlign error = errors.New("data from lower granularities does not align")
)
// Each metric on each level has it's own buffer.
// This is where the actual values go.
// If `cap(data)` is reached, a new buffer is created and
@ -80,18 +85,19 @@ func (b *buffer) write(ts int64, value Float) (*buffer, error) {
// the actual `from`/`to`.
// This function goes back the buffer chain if `from` is older than the
// currents buffer start.
func (b *buffer) read(from, to int64) ([]Float, int64, int64, error) {
// The loaded values are added to `data` and `data` is returned, possibly with a shorter length.
// If `data` is not long enough to hold all values, this function will panic!
func (b *buffer) read(from, to int64, data []Float) ([]Float, int64, int64, error) {
if from < b.start {
if b.prev != nil {
return b.prev.read(from, to)
return b.prev.read(from, to, data)
}
from = b.start
}
data := make([]Float, 0, (to-from)/b.frequency+1)
var t int64
for t = from; t < to; t += b.frequency {
var i int = 0
var t int64 = from
for ; t < to; t += b.frequency {
idx := int((t - b.start) / b.frequency)
if idx >= cap(b.data) {
b = b.next
@ -102,13 +108,44 @@ func (b *buffer) read(from, to int64) ([]Float, int64, int64, error) {
}
if t < b.start || idx >= len(b.data) {
data = append(data, NaN)
data[i] += NaN
} else {
data = append(data, b.data[idx])
data[i] += b.data[idx]
}
i++
}
return data, from, t, nil
return data[:i], from, t, nil
}
// Free up and free all buffers in the chain only containing data
// older than `t`.
func (b *buffer) free(t int64) (int, error) {
end := b.start + int64(len(b.data))*b.frequency
if end < t && b.next != nil {
b.next.prev = nil
n := 0
for b != nil {
prev := b.prev
if prev != nil && prev.start > b.start {
panic("time travel?")
}
n += 1
bufferPool.Put(b.data)
b.data = nil
b.next = nil
b.prev = nil
b = prev
}
return n, nil
}
if b.prev != nil {
return b.prev.free(t)
}
return 0, nil
}
// Could also be called "node" as this forms a node in a tree structure.
@ -158,77 +195,82 @@ func (l *level) findLevelOrCreate(selector []string) *level {
}
// This function assmumes that `l.lock` is LOCKED!
// Read `buffer.read` for context. This function does
// a lot of short-lived allocations and copies if this is
// not the "native" level for the requested metric. There
// is a lot of optimization potential here!
// Read `buffer.read` for context.
// If this level does not have data for the requested metric, the data
// is aggregated timestep-wise from all the children (recursively).
// Optimization suggestion: Pass a buffer as argument onto which the values should be added.
func (l *level) read(metric string, from, to int64, aggregation string) ([]Float, int64, int64, error) {
func (l *level) read(metric string, from, to int64, data []Float) ([]Float, int, int64, int64, error) {
if b, ok := l.metrics[metric]; ok {
// Whoo, this is the "native" level of this metric:
return b.read(from, to)
data, from, to, err := b.read(from, to, data)
return data, 1, from, to, err
}
if len(l.children) == 0 {
return nil, 0, 0, errors.New("no data for that metric/level")
return nil, 1, 0, 0, ErrNoData
}
if len(l.children) == 1 {
for _, child := range l.children {
child.lock.Lock()
data, from, to, err := child.read(metric, from, to, aggregation)
child.lock.Unlock()
return data, from, to, err
}
}
// "slow" case: We need to accumulate metrics accross levels/scopes/tags/whatever.
var data []Float = nil
n := 0
for _, child := range l.children {
child.lock.Lock()
cdata, cfrom, cto, err := child.read(metric, from, to, aggregation)
child.lock.Unlock()
child.lock.RLock()
cdata, cn, cfrom, cto, err := child.read(metric, from, to, data)
child.lock.RUnlock()
if err == ErrNoData {
continue
}
if err != nil {
return nil, 0, 0, err
return nil, 0, 0, 0, err
}
if data == nil {
if n == 0 {
data = cdata
from = cfrom
to = cto
n += cn
continue
}
if cfrom != from || cto != to {
// TODO: Here, we could take the max of cfrom and from and the min of cto and to instead.
// This would mean that we also have to resize data.
return nil, 0, 0, errors.New("data for metrics at child levels does not align")
return nil, 0, 0, 0, ErrDataDoesNotAlign
}
if len(data) != len(cdata) {
panic("WTF? Different freq. at different levels?")
}
for i := 0; i < len(data); i++ {
data[i] += cdata[i]
n += cn
}
if n == 0 {
return nil, 0, 0, 0, ErrNoData
}
return data, n, from, to, nil
}
func (l *level) free(t int64) (int, error) {
l.lock.Lock()
defer l.lock.Unlock()
n := 0
for _, b := range l.metrics {
m, err := b.free(t)
n += m
if err != nil {
return n, err
}
}
switch aggregation {
case "sum":
return data, from, to, nil
case "avg":
normalize := 1. / Float(len(l.children))
for i := 0; i < len(data); i++ {
data[i] *= normalize
for _, l := range l.children {
m, err := l.free(t)
n += m
if err != nil {
return n, err
}
return data, from, to, nil
default:
return nil, 0, 0, errors.New("invalid aggregation strategy: " + aggregation)
}
return n, nil
}
type MemoryStore struct {
@ -258,7 +300,8 @@ func (m *MemoryStore) Write(selector []string, ts int64, metrics []Metric) error
if !ok {
minfo, ok := m.metrics[metric.Name]
if !ok {
return errors.New("unkown metric: " + metric.Name)
// return errors.New("unkown metric: " + metric.Name)
continue
}
// First write to this metric and level
@ -279,6 +322,9 @@ func (m *MemoryStore) Write(selector []string, ts int64, metrics []Metric) error
return nil
}
// Returns all values for metric `metric` from `from` to `to` for the selected level.
// If the level does not hold the metric itself, the data will be aggregated recursively from the children.
// See `level.read` for more information.
func (m *MemoryStore) Read(selector []string, metric string, from, to int64) ([]Float, int64, int64, error) {
l := m.root.findLevelOrCreate(selector)
l.lock.RLock()
@ -293,5 +339,28 @@ func (m *MemoryStore) Read(selector []string, metric string, from, to int64) ([]
return nil, 0, 0, errors.New("unkown metric: " + metric)
}
return l.read(metric, from, to, minfo.Aggregation)
data := make([]Float, (to-from)/minfo.Frequency)
data, n, from, to, err := l.read(metric, from, to, data)
if err != nil {
return nil, 0, 0, err
}
if n > 1 {
if minfo.Aggregation == "avg" {
normalize := 1. / Float(n)
for i := 0; i < len(data); i++ {
data[i] *= normalize
}
} else if minfo.Aggregation != "sum" {
return nil, 0, 0, errors.New("invalid aggregation strategy: " + minfo.Aggregation)
}
}
return data, from, to, err
}
// Release all buffers for the selected level and all its children that contain only
// values older than `t`.
func (m *MemoryStore) Free(selector []string, t int64) (int, error) {
return m.root.findLevelOrCreate(selector).free(t)
}

View File

@ -302,6 +302,56 @@ func TestMemoryStoreArchive(t *testing.T) {
}
}
func TestMemoryStoreFree(t *testing.T) {
store := NewMemoryStore(map[string]MetricConfig{
"a": {Frequency: 1},
"b": {Frequency: 2},
})
count := 3000
sel := []string{"cluster", "host", "1"}
for i := 0; i < count; i++ {
err := store.Write(sel, int64(i), []Metric{
{Name: "a", Value: Float(i)},
{Name: "b", Value: Float(i)},
})
if err != nil {
t.Fatal(err)
}
}
n, err := store.Free([]string{"cluster", "host"}, int64(BUFFER_CAP*2)+100)
if err != nil {
t.Fatal(err)
}
if n != 3 {
t.Fatal("two buffers expected to be released")
}
adata, from, to, err := store.Read([]string{"cluster", "host", "1"}, "a", 0, int64(count))
if err != nil {
t.Fatal(err)
}
if from != int64(BUFFER_CAP*2) || to != int64(count) || len(adata) != count-2*BUFFER_CAP {
t.Fatalf("unexpected values from call to `Read`: from=%d, to=%d, len=%d", from, to, len(adata))
}
bdata, from, to, err := store.Read([]string{"cluster", "host", "1"}, "b", 0, int64(count))
if err != nil {
t.Fatal(err)
}
if from != int64(BUFFER_CAP*2) || to != int64(count) || len(bdata) != (count-2*BUFFER_CAP)/2 {
t.Fatalf("unexpected values from call to `Read`: from=%d, to=%d, len=%d", from, to, len(bdata))
}
if adata[0] != Float(BUFFER_CAP*2) || adata[len(adata)-1] != Float(count-1) {
t.Fatal("wrong values")
}
}
func BenchmarkMemoryStoreConcurrentWrites(b *testing.B) {
frequency := int64(5)
count := b.N