Distributed counters pattern in Firebase - Step-by-Step Execution

Process Flow - Distributed counters pattern

Start increment request

↓

Select a shard randomly

↓

Increment counter in selected shard

↓

Save updated shard count

↓

Repeat increments as needed

↓

To get total count

↓

Read all shards

↓

Sum all shard counts

↓

Return total count

Each increment updates one shard randomly to reduce contention. Total count is sum of all shards.

Execution Sample

Firebase

function incrementCounter() {
  const shardId = randomShard();
  const shardRef = db.collection('counters').doc(shardId);
  shardRef.update({count: firebase.firestore.FieldValue.increment(1)});
}

function getTotalCount() {
  return db.collection('counters').get().then(sumShards); 
}

Increment updates one shard; total count sums all shards.

Process Table

Step	Action	Shard Selected	Shard Count Before	Shard Count After	Total Count Calculation
1	Increment request received	-	-	-	-
2	Select shard randomly	shard_3	-	-	-
3	Read shard_3 count	shard_3	5	-	-
4	Increment shard_3 count	shard_3	5	6	-
5	Save updated shard_3 count	shard_3	-	6	-
6	Increment request received	-	-	-	-
7	Select shard randomly	shard_1	-	-	-
8	Read shard_1 count	shard_1	3	-	-
9	Increment shard_1 count	shard_1	3	4	-
10	Save updated shard_1 count	shard_1	-	4	-
11	Request total count	-	-	-	-
12	Read all shards counts	-	-	-	shard_1=4, shard_2=2, shard_3=6
13	Sum all shards	-	-	-	4 + 2 + 6 = 12
14	Return total count	-	-	-	12

💡 Total count returned after summing all shard counts.

Status Tracker

Variable	Start	After Step 4	After Step 9	After Step 13
shard_1_count	3	3	4	4
shard_2_count	2	2	2	2
shard_3_count	5	6	6	6
total_count	0	0	0	12

Key Moments - 3 Insights

Why do we update only one shard per increment instead of all shards?

Updating one shard reduces conflicts and improves performance, as shown in steps 2-5 and 7-10 where only one shard is incremented each time.

How is the total count calculated from shards?

The total count is the sum of all shard counts, as shown in steps 12-14 where all shards are read and summed.

What happens if two increments select the same shard at the same time?

Firebase handles concurrent updates safely with atomic increments, so counts are correctly updated without loss, as implied in steps 3-5 and 8-10.

Visual Quiz - 3 Questions

Look at the execution table, what is the count of shard_3 after step 4?

A 5

B 4

C 6

D 3

💡 Hint

Check the 'Shard Count After' column at step 4 in the execution_table.

At which step does the total count become available?

A Step 14

B Step 12

C Step 11

D Step 10

💡 Hint

Look at the 'Total Count Calculation' column in the execution_table for when the total is returned.

If increments always selected the same shard, what would change in the execution_table?

A Total count would not change

B Only one shard's count would increase repeatedly

C All shards would increase evenly

D Shards would be deleted

💡 Hint

Refer to the 'Shard Selected' column in the execution_table to see shard selection impact.

Concept Snapshot

Distributed counters split count into shards.
Each increment updates one shard randomly.
Total count is sum of all shards.
Reduces write conflicts and scales better.
Use atomic increments for safe updates.

Full Transcript

The distributed counters pattern splits a counter into multiple shards to reduce write conflicts. When an increment request happens, the system picks one shard randomly and increments its count atomically. This avoids many clients trying to update the same document at once. To get the total count, the system reads all shards and sums their counts. This pattern improves performance and scalability in Firebase by distributing load. The execution table shows increments updating individual shards and the total count calculated by summing all shards. Variables track shard counts changing step-by-step. Key moments clarify why only one shard is updated per increment and how totals are computed. The visual quiz tests understanding of shard counts and total calculation steps.