What is the advantage of GZIP vs DEFLATE compression?

Question

I have a web site in asp.NET 4 (C#).

I’m trying to find a way to better optimize bandwidth for my website.

I read many articles saying that DEFLATE is faster and smaller that GZIP because GZIP (based on DEFLATE) adds some extra data.

Checking the headers of bing.com and google.com it seems that they both send GZIP-encoded data.

Assuming what I read is true, I miss the advantage of GZIP in this case. So I suspect there should be a good reason to prefer GZIP to DEFLATE.

My questions:

• Does GZIP offer any advantage over DEFLATE I'm not aware of?
• Any clue why major search engines use GZIP?

Here’s the code I’m using to send DEFLATE (from Global.asax):

protected void Application_PreRequestHandlerExecute(object sender, EventArgs e)
    {

        HttpApplication app = sender as HttpApplication;
        string acceptEncoding = app.Request.Headers["Accept-Encoding"];
        Stream prevUncompressedStream = app.Response.Filter;

        if (!(app.Context.CurrentHandler is Page ||
            app.Context.CurrentHandler.GetType().Name == "SyncSessionlessHandler") ||
            app.Request["HTTP_X_MICROSOFTAJAX"] != null)
            return;

        if (acceptEncoding == null || acceptEncoding.Length == 0)
            return;

        acceptEncoding = acceptEncoding.ToLower();

        if (acceptEncoding.Contains("deflate") || acceptEncoding == "*")
        {
            // defalte
            app.Response.Filter = new DeflateStream(prevUncompressedStream,
                CompressionMode.Compress);
            app.Response.AppendHeader("Content-Encoding", "deflate");
        }
        else if (acceptEncoding.Contains("gzip"))
        {
            // gzip
            app.Response.Filter = new GZipStream(prevUncompressedStream,
                CompressionMode.Compress);
            app.Response.AppendHeader("Content-Encoding", "gzip");
        }
    }

Solution 1

Gzip is the more reliable because it is deflate plus a few headers and a check sum. In other words gzip is deflate, and extra headers and check sum. Deflate is checked with adler32, which is also part of gzip. Because the gzip payload is a DEFLATE-compressed payload.

Deflate info

Gzip info

a gzip file/stream contains:

- a 10-byte header, containing a magic number, a version number and a time stamp
- optional extra headers, such as the original file name,
- a body, containing a DEFLATE-compressed payload
- an 8-byte footer, containing a CRC-32 checksum and the length of the original uncompressed data

Solution 2

The other answer is mostly wrong. The "deflate" value for the Content-Encoding HTTP header is a misnomer that actually means ZLIB. Given that, both have checksums and the same compressed content. They only differ in their headers/footer and which checksum they use.

	gzip	"deflate" (zlib)
Header size	10 bytes	2 bytes
Footer size	4 bytes	0
Checksum	CRC32	Adler-32
Compression algorithm	DEFLATE	DEFLATE
Specification	RFC1952	RFC1950

Historically, "deflate" was problematic because early Microsoft IIS servers would send raw deflate data instead of zlib data (see https://stackoverflow.com/a/9186091/1218408). To work around that, it became common to just use gzip.

Deflate is very slightly faster because Adler-32 is typically faster to compute than CRC32, but the majority of the time is spent actually compressing data and not calculating the checksum.

Comments

[Sebastian Paaske Tørholm]

See stackoverflow.com/questions/388595/…

[GibboK]

Thanks Paul for your revision.

[Pacerier]

All answers are wrong / incomplete, except this: stackoverflow.com/a/9856879/632951

[GibboK]

To summarise.. do you mean GZIP is more browser compatible?

[Peter]

GZip will enable the browser to check for errors, deflate won't

[Pacerier]

@peer This is wrong, deflate will check for errors using adler32.

[Peter]

@Pacerier you are right the DEFLATE payload is checked by a checksum.

[david_adler]

don't you mean adler-32 is slight faster than crc32?

[Phil P]

I think that you've confused DEFLATE for ZLIB. ZLIB is RFC1950, your reference, and mostly matches your description. DEFLATE is RFC1951.

[ZachB]

@PhilP I didn't. I used the word "deflate" because that's the ill-named value of the Content-Encoding header in the HTTP spec. You're right that it's actually ZLIB, and therein lies the problem I discussed with Microsoft IIS servers. I edited the answer to emphasize this.