use memory based cost model for events cache

session/session.go

@@ -38,7 +38,7 @@ func New(mem memory.Allocator, resource string, storage db.Storage,
 	primary index.Primary, opts ...lsm.Option) *Session {
 	cache, err := ristretto.NewCache(&ristretto.Config{
 		NumCounters: 1e7,
-		MaxCost:     2 << 20,
+		MaxCost:     100 << 20, // 100MiB
 		BufferItems: 64,
 		OnEvict: func(item *ristretto.Item) {
 			item.Value.(*staples.Event).Release()
@@ -98,7 +98,7 @@ func (s *Session) process(ctx context.Context, req *v1.Event) {
 	s.mu.Lock()
 	s.build.Append(e)
 	s.mu.Unlock()
-	s.cache.SetWithTTL(e.ID, e, 1, DefaultSession)
+	s.cache.SetWithTTL(e.ID, e, int64(e.Size()), DefaultSession)
 }
 
 func (s *Session) Scan(ctx context.Context, start, end int64, fs *v1.Filters) (arrow.Record, error) {

staples/event.go

@@ -7,6 +7,7 @@ import (
 	"strings"
 	"sync"
 	"time"
+	"unsafe"
 
 	"github.com/cespare/xxhash/v2"
 	v1 "github.com/vinceanalytics/vince/gen/go/staples/v1"
@@ -56,6 +57,36 @@ type Event struct {
 	UtmTerm        string
 }
 
+// Size without strings
+var baseSize = unsafe.Sizeof(Event{})
+
+// Size in bytes of e in memory. We use this as a cost to control cache size.
+func (e *Event) Size() (n int) {
+	n = int(baseSize)
+	n += len(e.Browser)
+	n += len(e.BrowserVersion)
+	n += len(e.City)
+	n += len(e.Country)
+	n += len(e.Domain)
+	n += len(e.EntryPage)
+	n += len(e.ExitPage)
+	n += len(e.Host)
+	n += len(e.Event)
+	n += len(e.Os)
+	n += len(e.OsVersion)
+	n += len(e.Path)
+	n += len(e.Referrer)
+	n += len(e.ReferrerSource)
+	n += len(e.Region)
+	n += len(e.Screen)
+	n += len(e.UtmCampaign)
+	n += len(e.UtmContent)
+	n += len(e.UtmMedium)
+	n += len(e.UtmSource)
+	n += len(e.UtmTerm)
+	return
+}
+
 var eventsPool = &sync.Pool{New: func() any { return new(Event) }}
 
 func NewEvent() *Event {