SJF4J vs Jayway JsonPath: Up to 7x Faster in Java Benchmarks

We benchmarked JSONPath in Java.

SJF4J: https://github.com/sjf4j-projects/sjf4j

Jayway JsonPath: https://github.com/json-path/JsonPath

Result:

SJF4J is up to 7x faster.

Here’s what we found.

---

Benchmark Setup

We built a JMH benchmark suite comparing:

• SJF4J v1.5.0
• Jayway JsonPath v2.9.0

Across three workload shapes:

• Path compilation
• Query on shared JsonNode
• Query on Map/List object graphs

JMH config:

• 5 × 300ms warmup
• 8 × 300ms measurement
• 2 forks
• 1 thread

Expressions used:

• $.store.book[1].price
• $.store.bicycle.color
• $.store.book[*].author
• $..price
• $.store.book[?(@.price > 10)].title
• $.store.book[0,2].title

---

Headline Results

Geometric mean (lower is better):

Benchmark group	SJF4J	Jayway	Result
compile	97 ns/op	125 ns/op	SJF4J 1.28x faster
query_definite	100 ns/op	237 ns/op	SJF4J 2.38x faster
query_indefinite	656 ns/op	1294 ns/op	SJF4J 1.97x faster
query_map_list_definite	31 ns/op	200 ns/op	SJF4J 6.38x faster
query_map_list_indefinite	280 ns/op	1301 ns/op	SJF4J 4.65x faster

TL;DR

• ~2x faster on typical queries
• up to 7x faster on Map/List
• faster compilation as well

---

1) Query on Shared JsonNode

This is the fairest comparison: both libraries query the same parsed tree.

Definite paths

Expression	SJF4J	Jayway	Result
$.store.book[1].price	111 ns/op	260 ns/op	2.35x faster
$.store.bicycle.color	90 ns/op	216 ns/op	2.40x faster

Indefinite paths

Expression	SJF4J	Jayway	Result
$.store.book[*].author	342 ns/op	910 ns/op	2.66x faster
$..price	1736 ns/op	3877 ns/op	2.23x faster
$.store.book[?(@.price > 10)].title	1292 ns/op	1660 ns/op	1.29x faster
$.store.book[0,2].title	242 ns/op	478 ns/op	1.98x faster

Conclusion:

SJF4J is consistently faster as a JSONPath engine.

---

🚀 2) Query on Map/List Object Graphs

This is where the gap becomes much larger.

Both libraries support Map/List style object graphs, so this is still a fair comparison.

Definite paths

Expression	SJF4J	Jayway	Result
$.store.book[1].price	37 ns/op	280 ns/op	7.50x faster
$.store.bicycle.color	26 ns/op	142 ns/op	5.42x faster

Indefinite paths

Expression	SJF4J	Jayway	Result
$.store.book[*].author	135 ns/op	1045 ns/op	7.73x faster
$..price	543 ns/op	2495 ns/op	4.59x faster
$.store.book[?(@.price > 10)].title	823 ns/op	1836 ns/op	2.23x faster
$.store.book[0,2].title	102 ns/op	598 ns/op	5.89x faster

Takeaway:

If your application already works on Map/List object graphs,

this is the strongest result in the entire benchmark.

---

🔥 3) Native Object Graphs: POJO / JOJO

This is where SJF4J goes beyond traditional JSONPath libraries.

Instead of forcing everything into a JSON AST, SJF4J runs directly on:

• Map / List
• JOJO (JSON Object + Java Object hybrid)
• POJO

Jayway does not support this model natively.

So here we compare SJF4J across different object representations.

Geometric mean

Benchmark group	Map/List	JOJO	POJO
definite	31 ns/op	43 ns/op	94 ns/op
indefinite	270 ns/op	373 ns/op	554 ns/op

Representative results

Expression	Map/List	JOJO	POJO
$.store.book[1].price	37 ns/op	46 ns/op	102 ns/op
$.store.bicycle.color	26 ns/op	40 ns/op	87 ns/op
$.store.book[*].author	131 ns/op	196 ns/op	374 ns/op
$..price	496 ns/op	914 ns/op	955 ns/op
$.store.book[?(@.price > 10)].title	807 ns/op	876 ns/op	1100 ns/op

What this shows:

• Map/List is fastest (no abstraction)
• JOJO adds minimal overhead
• POJO is slower, but still practical

The key insight:

You can run JSONPath directly on your domain model.

No conversion. No intermediate tree.

---

What This Means in Practice

Instead of:

POJO → JsonNode → JSONPath → result

You can do:

POJO → JSONPath → result

This removes an entire layer from your data processing pipeline.

---

Why Is SJF4J Faster?

The performance difference mainly comes from:

• no forced conversion into a dedicated AST
• direct execution on native Java object graphs
• lower abstraction overhead in path evaluation

In short:

SJF4J avoids work that other libraries must do first.

---

Bottom Line

Jayway is fine for basic JSONPath usage.

But if performance matters, especially on Map/List or POJO object graphs:

SJF4J is simply faster.

---

Links

GitHub: https://github.com/sjf4j-projects/sjf4j

Benchmark source: JsonPathCompareBenchmark.java