---
title: "Aggregating high-volume events upstream (sessioning)"
slug: "aggregating-high-volume-events-upstream-sessioning"
updated: 2026-06-05T12:55:51Z
published: 2026-06-05T12:55:51Z
canonical: "docs.xtremepush.com/aggregating-high-volume-events-upstream-sessioning"
---

> ## Documentation Index
> Fetch the complete documentation index at: https://docs.xtremepush.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Aggregating high-volume events upstream (sessioning)

This page provides information on managing high-volume events upstream and implementation examples for each supported managed Kafka platform.

## Types of Aggregation

### Session-based aggregates

Capped by time (e.g. ≤10 minutes) or count (e.g. ≤N spins). Each aggregate should include totals for stakes, wins, net, counts, game code, channel, and start/end timestamps.

### Rolling window aggregates

Typically 2–5 minute summaries grouped by user or product.

> [!NOTE]
> **Why** `EMIT FINAL` **/** `DISCARDING` **accumulation matters**
> 
> Session window aggregations can fire multiple times as new events arrive within an open session. If your pipeline uses the default incremental-update behaviour, downstream systems receive multiple partial records for the same session — causing event counts to be inflated. The `EMIT FINAL` (ksqlDB), closed-window-only emission (Kafka Streams), and `AccumulationMode.DISCARDING` (Beam) patterns each guarantee that **exactly one record is emitted per completed session**, which is the correct input for Xtremepush event-level analytics.

### **Confluent Cloud (ksqlDB)**

**Session aggregation**

```sql
CREATE TABLE session_summaries
  WITH (KAFKA_TOPIC = 'xp-session-events', VALUE_FORMAT = 'JSON')
AS SELECT
  user_id,
  'session'                                          AS event,
  AS_VALUE(user_id)                                  AS user_id,
  TIMESTAMPTOSTRING(MIN(ROWTIME), 'yyyy-MM-dd HH:mm:ss', 'UTC') AS `timestamp`,
  STRUCT(
    session_start_ms := MIN(ROWTIME),
    session_end_ms   := MAX(ROWTIME),
    event_count      := COUNT(*),
    duration_ms      := MAX(ROWTIME) - MIN(ROWTIME)
  )                                                  AS value,
  STRUCT()                                           AS user_attributes
FROM xtremepush_events
WINDOW SESSION (10 MINUTES)
GROUP BY user_id
EMIT FINAL;
```

`EMIT FINAL` **is critical.** Without it, ksqlDB emits incremental updates as the session is being built. `EMIT FINAL` ensures exactly one record is emitted per closed session (after the inactivity gap expires), keeping downstream event counts meaningful.

**Rolling window (2-minute summary)**

```sql
CREATE TABLE rolling_2min_summary
  WITH (KAFKA_TOPIC = 'xp-rolling-2min', VALUE_FORMAT = 'JSON')
AS SELECT
  user_id,
  'session_summary'                                  AS event,
  AS_VALUE(user_id)                                  AS user_id,
  TIMESTAMPTOSTRING(WINDOWSTART, 'yyyy-MM-dd HH:mm:ss', 'UTC') AS `timestamp`,
  STRUCT(
    window_start_ms := WINDOWSTART,
    window_end_ms   := WINDOWEND,
    event_count     := COUNT(*)
  )                                                  AS value,
  STRUCT()                                           AS user_attributes
FROM xtremepush_events
WINDOW TUMBLING (SIZE 2 MINUTES)
GROUP BY user_id
EMIT FINAL;
```

### **Amazon MSK (Kafka Streams — Java)**

**Session aggregation**

```java
import org.apache.kafka.streams.kstream.SessionWindows;
import java.time.Duration;

KTable<Windowed<String>, Long> sessions = rawStream
    .groupByKey()
    .windowedBy(
        SessionWindows.ofInactivityGapAndGrace(
            Duration.ofMinutes(10),   // inactivity gap — closes session after 10min silence
            Duration.ofMinutes(2)     // grace period — accept late-arriving events
        )
    )
    .aggregate(
        () -> 0L,                                      // initialiser
        (key, value, aggregate) -> aggregate + 1,      // incrementor
        (key, agg1, agg2) -> agg1 + agg2              // merger (called when two sessions merge)
    );

sessions.toStream().map((windowed, count) -> {
    try {
        ObjectNode envelope = mapper.createObjectNode();
        envelope.put("event",   "session");
        envelope.put("user_id", windowed.key());

        String ts = Instant.ofEpochMilli(windowed.window().end())
            .toString().replace("T", " ").replaceAll("\\.\\d+Z$", "");
        envelope.put("timestamp", ts);

        ObjectNode value = mapper.createObjectNode();
        value.put("session_start_ms", windowed.window().start());
        value.put("session_end_ms",   windowed.window().end());
        value.put("event_count",      count);
        value.put("duration_ms",      windowed.window().end() - windowed.window().start());
        envelope.set("value", value);
        envelope.set("user_attributes", mapper.createObjectNode());

        return KeyValue.pair(windowed.key(), mapper.writeValueAsString(envelope));
    } catch (Exception e) {
        throw new RuntimeException(e);
    }
}).to("xp-session-events");
```

The **merger** function `(key, agg1, agg2) -&gt; agg1 + agg2` is required for session windows. Kafka Streams may merge two adjacent session windows when a late-arriving event bridges the gap between them — the merger combines the counts from both windows.

**Rolling window (2-minute summary)**

```java
import org.apache.kafka.streams.kstream.TimeWindows;

KTable<Windowed<String>, Long> rolling = rawStream
    .groupByKey()
    .windowedBy(
        TimeWindows.ofSizeAndGrace(Duration.ofMinutes(2), Duration.ofMinutes(1))
    )
    .count();
```

### **Google Cloud Managed Apache Kafka (Apache Beam / Dataflow)**

**Session aggregation**

```python
import apache_beam as beam
from apache_beam.transforms.window import Sessions
from apache_beam.transforms.trigger import AfterWatermark, AccumulationMode
import json
from datetime import datetime, timezone

class SessionOutputFn(beam.DoFn):
    def process(self, element, window=beam.DoFn.WindowParam):
        user_id, events = element
        ts = datetime.fromtimestamp(window.end, tz=timezone.utc).strftime('%Y-%m-%d %H:%M:%S')
        yield json.dumps({
            'event':   'session',
            'user_id': user_id,
            'timestamp': ts,
            'value': {
                'session_start_ms': int(window.start * 1000),
                'session_end_ms':   int(window.end   * 1000),
                'event_count':      len(list(events)),
                'duration_ms':      int((window.end - window.start) * 1000),
            },
            'user_attributes': {}
        })

(
    events_pcoll
    | 'ExtractUserId'  >> beam.Map(lambda e: (json.loads(e)['user_id'], e))
    | 'SessionWindow'  >> beam.WindowInto(
                              Sessions(gap_size=600),        # 10-minute inactivity gap
                              trigger=AfterWatermark(),
                              accumulation_mode=AccumulationMode.DISCARDING)
    | 'GroupByKey'     >> beam.GroupByKey()
    | 'BuildSession'   >> beam.ParDo(SessionOutputFn())
    | 'WriteToKafka'   >> beam.io.WriteToKafka(
                              producer_config={'bootstrap.servers': 'BROKER:9092'},
                              topic='xp-session-events')
)
```

`AccumulationMode.DISCARDING` **is critical.** Beam windows fire when the watermark passes the end of the window. With `DISCARDING`, each pane discards previously accumulated data after firing — so exactly one complete session record is emitted per closed session. Without this (`ACCUMULATING` mode), you would receive incremental partial-session updates that would overcount events downstream.

**Rolling window (2-minute summary)**

```python
from apache_beam.transforms.window import FixedWindows
from apache_beam.transforms.trigger import AfterWatermark, AccumulationMode

(
    events_pcoll
    | 'KeyByUser'     >> beam.Map(lambda e: (json.loads(e)['user_id'], 1))
    | 'FixedWindow'   >> beam.WindowInto(
                             FixedWindows(120),              # 2-minute fixed windows
                             trigger=AfterWatermark(),
                             accumulation_mode=AccumulationMode.DISCARDING)
    | 'CountPerUser'  >> beam.CombinePerKey(sum)
    | 'BuildSummary'  >> beam.Map(lambda kv: json.dumps({
                             'event':   'session_summary',
                             'user_id': kv[0],
                             'timestamp': datetime.utcnow().strftime('%Y-%m-%d %H:%M:%S'),
                             'value':   {'event_count': kv[1]},
                             'user_attributes': {}
                         }))
    | 'WriteToKafka'  >> beam.io.WriteToKafka(...)
)
```