Glen Mazza's Weblog

Compressing SOAP messages during transit

In this article I'll be demonstrating two ways to compress Apache CXF SOAP requests and responses to conserve network resources--Fast Infoset (FI) and GZip compression. We'll modify the introductory DoubleIt web service tutorial to test this functionality along with Wireshark to see how the SOAP requests and responses change as a result. The code modifications necessary are quite minimal, just a few lines over one or two source files.

As a reference point, let's use Wireshark to see the SOAP request and response for a standard DoubleIt call without compression:

POST /doubleit/services/doubleit HTTP/1.1
Content-Type: text/xml; charset=UTF-8
Accept: */*
SOAPAction: ""
User-Agent: Apache-CXF/3.2.2
Cache-Control: no-cache
Pragma: no-cache
Host: localhost:8080
Connection: keep-alive
Content-Length: 224

<soap:Envelope xmlns:soap="http://schemas.xmlsoap.org/soap/envelope/">
    <soap:Body>
        <ns2:DoubleIt xmlns:ns2="http://www.example.org/schema/DoubleIt">
            <numberToDouble>10</numberToDouble>
        </ns2:DoubleIt>
    </soap:Body>
</soap:Envelope>
        
HTTP/1.1 200
Content-Type: text/xml;charset=UTF-8
Content-Length: 238
Date: Sat, 10 Mar 2018 16:09:14 GMT

<soap:Envelope xmlns:soap="http://schemas.xmlsoap.org/soap/envelope/">
    <soap:Body>
        <ns2:DoubleItResponse xmlns:ns2="http://www.example.org/schema/DoubleIt">
            <doubledNumber>20</doubledNumber>
        </ns2:DoubleItResponse>
    </soap:Body>
</soap:Envelope>

First, let's activate fast infoset, both client- and service-side. The negotiation process involves the SOAP client sending an uncompressed Accept: application/fastinfoset HTTP header in the first SOAP request. If the server is configured to support FI, it will return a compressed SOAP response with a content type of application/fastinfoset. Subsequently all SOAP requests and responses between the client and web service provider will then use FI. However, if the server is not configured to use FI the server will return an uncompressed SOAP response with the usual text/xml Content-Type. The following table shows how to activate FI:

@WebService(targetNamespace = "http://www.example.org/contract/DoubleIt"... 
@org.apache.cxf.annotations.FastInfoset
public class DoubleItPortTypeImpl implements DoubleItPortType {
    ....  
}

public class WSClient {

    public static void main (String[] args) {
        DoubleItService service = new DoubleItService();
        DoubleItPortType port = service.getDoubleItPort();

        Client client = ClientProxy.getClient(port);
        client.getInInterceptors().add(new org.apache.cxf.interceptor.FIStaxInInterceptor());
        client.getOutInterceptors().add(new org.apache.cxf.interceptor.FIStaxOutInterceptor());
        ...
    }
}

Now, let's check the results over the wire:

POST /doubleit/services/doubleit HTTP/1.1
Content-Type: text/xml; charset=UTF-8
Accept: application/fastinfoset, */*
SOAPAction: ""
User-Agent: Apache-CXF/3.2.2
Cache-Control: no-cache
Pragma: no-cache
Host: localhost:8080
Connection: keep-alive
Content-Length: 224

<soap:Envelope xmlns:soap="http://schemas.xmlsoap.org/soap/envelope/">
    <soap:Body>
        <ns2:DoubleIt xmlns:ns2="http://www.example.org/schema/DoubleIt">
            <numberToDouble>10</numberToDouble>
        </ns2:DoubleIt>
    </soap:Body>
</soap:Envelope>
        
HTTP/1.1 200
Content-Type: application/fastinfoset
Transfer-Encoding: chunked
Date: Sat, 10 Mar 2018 17:39:16 GMT

.....8..soap(http://schemas.xmlsoap.org/soap/envelope/.?...Envelope?...Body8..ns2%http://www.example.org/schema/DoubleIt.?...DoubleItResponse<.doubledNumber.20....POST /doubleit/services/doubleit HTTP/1.1
Content-Type: application/fastinfoset
Accept: application/fastinfoset, */*
SOAPAction: ""
User-Agent: Apache-CXF/3.2.2
Cache-Control: no-cache
Pragma: no-cache
Host: localhost:8080
Connection: keep-alive
Content-Length: 155

.....8..soap(http://schemas.xmlsoap.org/soap/envelope/.?...Envelope?...Body8..ns2%http://www.example.org/schema/DoubleIt.?...DoubleIt<
numberToDouble.0....HTTP/1.1 200
Content-Type: application/fastinfoset
Transfer-Encoding: chunked
Date: Sat, 10 Mar 2018 17:39:16 GMT

.....8..soap(http://schemas.xmlsoap.org/soap/envelope/.?...Envelope?...Body8..ns2%http://www.example.org/schema/DoubleIt.?...DoubleItResponse<.doubledNumber.0....

For GZIP, follow the below table for configuration information. The negotiation process involves the SOAP client sending an Accept-Encoding: gzip HTTP header in the first SOAP request. If the server is configured to support GZIP compression, it will return a compressed SOAP response with the Content-Encoding: gzip HTTP Header. Subsequently all SOAP requests and responses between the client and web service provider will then use GZIP. If the server is not configured to use GZIP then SOAP requests and responses will be sent as normal, uncompressed.

The results with GZIP are shown below. I set the threshold to zero, both client- and service-side, to ensure compression of all messages. (Note following Wireshark's HTTP stream will uncompress by default, causing one to think no compression is occurring. To see the compressed values as below, right-click any packet part of the SOAP calls and choose "Follow->TCPStream" instead.)

POST /doubleit/services/doubleit HTTP/1.1
Content-Type: text/xml; charset=UTF-8
Accept: */*
Accept-Encoding: gzip;q=1.0, identity; q=0.5, *;q=0
SOAPAction: ""
User-Agent: Apache-CXF/3.2.2
Cache-Control: no-cache
Pragma: no-cache
Host: localhost:8080
Connection: keep-alive
Content-Length: 224

<soap:Envelope xmlns:soap="http://schemas.xmlsoap.org/soap/envelope/">
    <soap:Body>
        <ns2:DoubleIt xmlns:ns2="http://www.example.org/schema/DoubleIt">
            <numberToDouble>10</numberToDouble>
        </ns2:DoubleIt>
    </soap:Body>
</soap:Envelope>HTTP/1.1 200
Content-Encoding: gzip
Vary: Accept-Encoding
Content-Type: text/xml;charset=UTF-8
Content-Length: 158
Date: Sat, 10 Mar 2018 18:03:20 GMT

        ..........m.... .D....X.#A.F.^<..T6..,..........d&.fW1.(.aFG.w.w.e..M?.Q...GoX.........)A.U...>.
        ...M...xC..x....M)    \.....:.k33m..:......l......s.l../&.......
        
POST /doubleit/services/doubleit HTTP/1.1
Content-Type: text/xml; charset=UTF-8
Accept: */*
Accept-Encoding: gzip;q=1.0, identity; q=0.5, *;q=0
Content-Encoding: gzip
SOAPAction: ""
User-Agent: Apache-CXF/3.2.2
Cache-Control: no-cache
Pragma: no-cache
Host: localhost:8080
Connection: keep-alive
Content-Length: 150

..........].1..0.E..z..1F!......`.!..:m..q..
.....l.\..8....)....c...4..;....*W.?...k..kf.../k"......5.>A_])E..B...f=|...Ch............u..?..7Y.......HTTP/1.1 200
Content-Encoding: gzip
Vary: Accept-Encoding
Content-Type: text/xml;charset=UTF-8
Content-Length: 157
Date: Sat, 10 Mar 2018 18:03:20 GMT

..........m.... .D....X.. .../...*.{..t.......x..d..*&..)<.Q...]`Y.}7..%.....E../h.C....tZ5...S..;y..wxNW.H..n...........l3.4..6i/..q.[.kC...b........Q.7....

Posted at 07:00AM Mar 10, 2018 by GlenMazza in Web Services | Comments[0]