Benchmark Results

Live benchmark comparisons between Dekaf and Confluent.Kafka, automatically updated on every commit to main.

Last Updated: 2026-03-22 15:57 UTC

info

These benchmarks run on GitHub Actions (ubuntu-latest) using BenchmarkDotNet. Ratio < 1.0 means Dekaf is faster than Confluent.Kafka

Producer Benchmarks

Comparing Dekaf vs Confluent.Kafka for message production across different scenarios.

Method	Categories	MessageSize	BatchSize	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Allocated	Alloc Ratio
Confluent_ProduceBatch	BatchProduce	100	100	5,913.64 μs	111.984 μs	74.071 μs	1.00	0.02	-	-	106.54 KB	1.00
Dekaf_ProduceBatch	BatchProduce	100	100	1,338.65 μs	149.387 μs	98.810 μs	0.23	0.02	-	-	35.83 KB	0.34
Confluent_ProduceBatch	BatchProduce	100	1000	7,242.95 μs	93.852 μs	62.077 μs	1.00	0.01	31.2500	15.6250	1062.81 KB	1.00
Dekaf_ProduceBatch	BatchProduce	100	1000	2,257.62 μs	57.221 μs	34.051 μs	0.31	0.01	15.6250	7.8125	488.02 KB	0.46
Confluent_ProduceBatch	BatchProduce	1000	100	6,468.26 μs	24.640 μs	14.663 μs	1.00	0.00	7.8125	-	194.04 KB	1.00
Dekaf_ProduceBatch	BatchProduce	1000	100	1,115.13 μs	27.604 μs	16.427 μs	0.17	0.00	-	-	39.28 KB	0.20
Confluent_ProduceBatch	BatchProduce	1000	1000	8,991.68 μs	1,046.170 μs	547.167 μs	1.00	0.08	78.1250	31.2500	1937.82 KB	1.00
Dekaf_ProduceBatch	BatchProduce	1000	1000	5,407.75 μs	484.469 μs	320.447 μs	0.60	0.05	46.8750	15.6250	1225.2 KB	0.63
Confluent_FireAndForget	FireAndForget	100	100	123.17 μs	5.228 μs	3.458 μs	1.00	0.04	1.7090	-	42.09 KB	1.00
Dekaf_FireAndForget	FireAndForget	100	100	58.69 μs	6.282 μs	4.155 μs	0.48	0.03	-	-	6.81 KB	0.16
Confluent_FireAndForget	FireAndForget	100	1000	1,261.30 μs	39.552 μs	26.161 μs	1.00	0.03	15.6250	-	421.32 KB	1.00
Dekaf_FireAndForget	FireAndForget	100	1000	678.08 μs	113.453 μs	67.514 μs	0.54	0.05	-	-	43.97 KB	0.10
Confluent_FireAndForget	FireAndForget	1000	100	NA	NA	NA	?	?	NA	NA	NA	?
Dekaf_FireAndForget	FireAndForget	1000	100	459.95 μs	26.742 μs	17.688 μs	?	?	-	-	8.37 KB	?
Confluent_FireAndForget	FireAndForget	1000	1000	NA	NA	NA	?	?	NA	NA	NA	?
Dekaf_FireAndForget	FireAndForget	1000	1000	4,306.57 μs	137.199 μs	81.645 μs	?	?	-	-	84.66 KB	?
Confluent_ProduceSingle	SingleProduce	100	100	5,366.03 μs	17.476 μs	11.560 μs	1.00	0.00	-	-	1.18 KB	1.00
Dekaf_ProduceSingle	SingleProduce	100	100	1,254.54 μs	25.299 μs	16.734 μs	0.23	0.00	-	-	1.61 KB	1.36
Confluent_ProduceSingle	SingleProduce	100	1000	5,368.40 μs	17.672 μs	11.689 μs	1.00	0.00	-	-	1.18 KB	1.00
Dekaf_ProduceSingle	SingleProduce	100	1000	1,236.80 μs	28.179 μs	18.639 μs	0.23	0.00	-	-	1.61 KB	1.36
Confluent_ProduceSingle	SingleProduce	1000	100	5,379.24 μs	10.845 μs	6.454 μs	1.00	0.00	-	-	2.06 KB	1.00
Dekaf_ProduceSingle	SingleProduce	1000	100	1,207.71 μs	33.761 μs	22.331 μs	0.22	0.00	-	-	1.61 KB	0.78
Confluent_ProduceSingle	SingleProduce	1000	1000	5,363.40 μs	8.927 μs	5.905 μs	1.00	0.00	-	-	2.06 KB	1.00
Dekaf_ProduceSingle	SingleProduce	1000	1000	1,146.42 μs	13.314 μs	8.806 μs	0.21	0.00	-	-	1.61 KB	0.78

Benchmarks with issues: ProducerBenchmarks.Confluent_FireAndForget: Job-CVXFWD(IterationCount=10, RunStrategy=Throughput, WarmupCount=3) [MessageSize=1000, BatchSize=100] ProducerBenchmarks.Confluent_FireAndForget: Job-CVXFWD(IterationCount=10, RunStrategy=Throughput, WarmupCount=3) [MessageSize=1000, BatchSize=1000]

Consumer Benchmarks

Comparing Dekaf vs Confluent.Kafka for message consumption.

Method	Categories	MessageCount	MessageSize	Mean	Error	StdDev	Ratio	Allocated	Alloc Ratio
Confluent_ConsumeAll	ConsumeAll	100	100	3.168 s	0.0034 s	0.0009 s	1.00	76592 B	1.00
Dekaf_ConsumeAll	ConsumeAll	100	100	3.011 s	0.0020 s	0.0005 s	0.95	133448 B	1.74
Confluent_ConsumeAll	ConsumeAll	100	1000	3.165 s	0.0026 s	0.0007 s	1.00	256544 B	1.00
Dekaf_ConsumeAll	ConsumeAll	100	1000	3.009 s	0.0045 s	0.0007 s	0.95	320856 B	1.25
Confluent_ConsumeAll	ConsumeAll	1000	100	3.165 s	0.0021 s	0.0006 s	1.00	616592 B	1.00
Dekaf_ConsumeAll	ConsumeAll	1000	100	3.010 s	0.0025 s	0.0006 s	0.95	542512 B	0.88
Confluent_ConsumeAll	ConsumeAll	1000	1000	3.165 s	0.0044 s	0.0012 s	1.00	2424608 B	1.00
Dekaf_ConsumeAll	ConsumeAll	1000	1000	3.010 s	0.0026 s	0.0004 s	0.95	2289088 B	0.94
Confluent_PollSingle	PollSingle	100	100	3.157 s	0.0015 s	0.0004 s	1.00	-	NA
Dekaf_PollSingle	PollSingle	100	100	3.006 s	0.0049 s	0.0008 s	0.95	31720 B	NA
Confluent_PollSingle	PollSingle	100	1000	3.156 s	0.0053 s	0.0014 s	1.00	-	NA
Dekaf_PollSingle	PollSingle	100	1000	3.006 s	0.0016 s	0.0002 s	0.95	34296 B	NA
Confluent_PollSingle	PollSingle	1000	100	3.157 s	0.0037 s	0.0010 s	1.00	-	NA
Dekaf_PollSingle	PollSingle	1000	100	3.006 s	0.0016 s	0.0004 s	0.95	32416 B	NA
Confluent_PollSingle	PollSingle	1000	1000	3.156 s	0.0035 s	0.0009 s	1.00	-	NA
Dekaf_PollSingle	PollSingle	1000	1000	3.006 s	0.0022 s	0.0006 s	0.95	36360 B	NA

Protocol Benchmarks

Zero-allocation wire protocol serialization/deserialization.

tip

Allocated = - means zero heap allocations - the goal of Dekaf's design!

Method	Mean	Error	StdDev	Allocated
'Write 1000 Int32s'	33.256 μs	13.0221 μs	7.7493 μs	-
'Write 100 Strings (100 chars)'	13.601 μs	0.2616 μs	0.1730 μs	-
'Write 100 CompactStrings'	11.224 μs	0.0978 μs	0.0511 μs	-
'Write 1000 VarInts'	36.555 μs	17.8534 μs	10.6243 μs	-
'Read 1000 Int32s'	16.521 μs	10.0349 μs	5.9716 μs	-
'Read 1000 VarInts'	29.695 μs	8.4714 μs	5.0412 μs	-
'Write RecordBatch (10 records)'	17.055 μs	0.4948 μs	0.2945 μs	-
'Read RecordBatch (10 records)'	4.007 μs	0.1558 μs	0.0815 μs	-

Serializer Benchmarks

Method	Mean	Error	StdDev	Ratio	RatioSD	Allocated	Alloc Ratio
'Serialize String (10 chars)'	1,433.1 ns	76.23 ns	50.42 ns	0.33	0.02	-	NA
'Serialize String (100 chars)'	1,551.0 ns	143.29 ns	94.78 ns	0.36	0.02	-	NA
'Serialize String (1000 chars)'	1,916.4 ns	414.49 ns	274.16 ns	0.44	0.06	-	NA
'Deserialize String'	2,874.9 ns	36.05 ns	21.45 ns	0.67	0.02	-	NA
'Serialize Int32'	713.0 ns	64.41 ns	38.33 ns	0.17	0.01	-	NA
'Serialize 100 Messages (key+value)'	32,273.4 ns	370.32 ns	220.37 ns	7.48	0.26	-	NA
'ArrayBufferWriter + Copy'	4,321.6 ns	269.47 ns	160.36 ns	1.00	0.05	-	NA
'PooledBufferWriter Direct'	3,662.2 ns	206.22 ns	122.72 ns	0.85	0.04	-	NA

Compression Benchmarks

Method	Mean	Error	StdDev	Allocated
'Snappy Compress 1KB'	11.210 μs	0.2835 μs	0.1875 μs	-
'Snappy Compress 1MB'	529.566 μs	23.2052 μs	15.3488 μs	-
'Snappy Decompress 1KB'	9.823 μs	0.3748 μs	0.2479 μs	-
'Snappy Decompress 1MB'	1,622.605 μs	30.4440 μs	18.1167 μs	-

---

How to Read These Results

• Mean: Average execution time
• Error: Half of 99.9% confidence interval
• StdDev: Standard deviation of all measurements
• Ratio: Performance relative to baseline (Confluent.Kafka)
  • < 1.0 = Dekaf is faster
  • > 1.0 = Confluent is faster
  • 1.0 = Same performance
• Allocated: Heap memory allocated per operation
  • - = Zero allocations (ideal!)

Benchmarks are automatically run on every push to main.