JSON and XML comparison
Processing speed may not be the only relevant matter, however, as that's the question, here are some numbers in a benchmark: JSON vs. XML: Some hard numbers about verbosity. For the speed, in this simple benchmark, XML presents a 21% overhead over JSON.
An important note about the verbosity, which is as the article says, the most common complain: this is not so much relevant in practice (neither XML nor JSON data are typically handled by humans, but by machines), even if for the matter of speed, it requires some reasonable more time to compress.
Also, in this benchmark, a big amount of data was processed, and a typical web application won't transmit data chunks of such sizes, as big as 90MB, and compression may not be beneficial (for small enough data chunks, a compressed chunk will be bigger than the uncompressed chunk), so not applicable.
Still, if no compression is involved, JSON, as obviously terser, will weight less over the transmission channel, especially if transmitted through a WebSocket connection, where the absence of the classic HTTP overhead may make the difference at the advantage of JSON, even more significant.
After transmission, data is to be consumed, and this count in the overall processing time. If big or complex enough data are to be transmitted, the lack of a schema automatically checked for by a validating XML parser, may require more check on JSON data; these checks would have to be executed in JavaScript, which is not known to be particularly fast, and so it may present an additional overhead over XML in such cases.
Anyway, only testing will provides the answer for your particular use‑case (if speed is really the only matter, and not standard nor safety nor integrity…).
Update 1: worth to mention, is EXI, the binary XML format, which offers compression at less cost than using Gzip, and save processing otherwise needed to decompress compressed XML. EXI is to XML, what BSON is to JSON. Have a quick overview here, with some reference to efficiency in both space and time: EXI: The last binary standard?.
Update 2: there also exists a binary XML performance reports, conducted by the W3C, as efficiency and low memory and CPU footprint, is also a matter for the XML area too: Efficient XML Interchange Evaluation.
Update 2015-03-01
Worth to be noticed in this context, as HTTP overhead was raised as an issue: the IANA has registered the EXI encoding (the efficient binary XML mentioned above), as a a Content Coding for the HTTP protocol (alongside with compress, deflate and gzip). This means EXI is an option which can be expected to be understood by browsers among possibly other HTTP clients. See Hypertext Transfer Protocol Parameters (iana.org).
Before answering when to use which one, a little background:
edit: I should mention that this comparison is really from the perspective of using them in a browser with JavaScript. It's not the way either data format has to be used, and there are plenty of good parsers which will change the details to make what I'm saying not quite valid.
JSON is both more compact and (in my view) more readable - in transmission it can be "faster" simply because less data is transferred.
In parsing, it depends on your parser. A parser turning the code (be it JSON or XML) into a data structure (like a map) may benefit from the strict nature of XML (XML Schemas disambiguate the data structure nicely) - however in JSON the type of an item (String/Number/Nested JSON Object) can be inferred syntactically, e.g:
myJSON = {"age" : 12,
"name" : "Danielle"}
The parser doesn't need to be intelligent enough to realise that 12
represents a number, (and Danielle
is a string like any other). So in javascript we can do:
anObject = JSON.parse(myJSON);
anObject.age === 12 // True
anObject.name == "Danielle" // True
anObject.age === "12" // False
In XML we'd have to do something like the following:
<person>
<age>12</age>
<name>Danielle</name>
</person>
(as an aside, this illustrates the point that XML is rather more verbose; a concern for data transmission). To use this data, we'd run it through a parser, then we'd have to call something like:
myObject = parseThatXMLPlease();
thePeople = myObject.getChildren("person");
thePerson = thePeople[0];
thePerson.getChildren("name")[0].value() == "Danielle" // True
thePerson.getChildren("age")[0].value() == "12" // True
Actually, a good parser might well type the age
for you (on the other hand, you might well not want it to). What's going on when we access this data is - instead of doing an attribute lookup like in the JSON example above - we're doing a map lookup on the key name
. It might be more intuitive to form the XML like this:
<person name="Danielle" age="12" />
But we'd still have to do map lookups to access our data:
myObject = parseThatXMLPlease();
age = myObject.getChildren("person")[0].getAttr("age");
EDIT: Original:
In most programming languages (not all, by any stretch) a map lookup such as this will be more costly than an attribute lookup (like we got above when we parsed the JSON).
This is misleading: remember that in JavaScript (and other dynamic languages) there's no difference between a map lookup and a field lookup. In fact, a field lookup is just a map lookup.
If you want a really worthwhile comparison, the best is to benchmark it - do the benchmarks in the context where you plan to use the data.
As I have been typing, Felix Kling has already put up a fairly succinct answer comparing them in terms of when to use each one, so I won't go on any further.
Faster is not an attribute of JSON or XML or a result that a comparison between those would yield. If any, then it is an attribute of the parsers or the bandwidth with which you transmit the data.
Here is (the beginning of) a list of advantages and disadvantages of JSON and XML:
JSON
Pro:
- Simple syntax, which results in less "markup" overhead compared to XML.
- Easy to use with JavaScript as the markup is a subset of JS object literal notation and has the same basic data types as JavaScript.
- JSON Schema for description and datatype and structure validation
- JsonPath for extracting information in deeply nested structures
Con:
Simple syntax, only a handful of different data types are supported.
No support for comments.
XML
Pro:
- Generalized markup; it is possible to create "dialects" for any kind of purpose
- XML Schema for datatype, structure validation. Makes it also possible to create new datatypes
- XSLT for transformation into different output formats
- XPath/XQuery for extracting information in deeply nested structures
- built in support for namespaces
Con:
- Relatively wordy compared to JSON (results in more data for the same amount of information).
So in the end you have to decide what you need. Obviously both formats have their legitimate use cases. If you are mostly going to use JavaScript then you should go with JSON.
Please feel free to add pros and cons. I'm not an XML expert ;)
I found this article at digital bazaar really interesting. Quoting their quotations from Norm:
About JSON pros:
If all you want to pass around are atomic values or lists or hashes of atomic values, JSON has many of the advantages of XML: it’s straightforwardly usable over the Internet, supports a wide variety of applications, it’s easy to write programs to process JSON, it has few optional features, it’s human-legible and reasonably clear, its design is formal and concise, JSON documents are easy to create, and it uses Unicode. ...
About XML pros:
XML deals remarkably well with the full richness of unstructured data. I’m not worried about the future of XML at all even if its death is gleefully celebrated by a cadre of web API designers.
And I can’t resist tucking an "I told you so!" token away in my desk. I look forward to seeing what the JSON folks do when they are asked to develop richer APIs. When they want to exchange less well strucured data, will they shoehorn it into JSON? I see occasional mentions of a schema language for JSON, will other languages follow? ...
I personally agree with Norm. I think that most attacks to XML come from Web Developers for typical applications, and not really from integration developers. But that's my opinion! ;)