Helping a colleague this week, we uncovered some odd behavior with a site whose performance he was analyzing. Upon first glance, it was clear that this site had a performance issue – they had HTTP persistence disabled. Immediate red flag in the areas of network overhead and geographic latency.

Further digging exposed something more sinister. It seems that HTTP persistence was only disabled for browsers with MSIE in the user-agent string. Even if the user-agent string was just MSIE, HTTP persistence was off.

The customer was very forthcoming and sent us their standard httpd.conf file. This showed no sign of the standard (and frustrating) global disabling of persistence for Internet Explorer.

Finally, it came to us. The customer had provided a simple network diagram, and there, just before packets hit the Internet, was a Layer 7 firewall. How did we know the Layer 7 firewall was the likely cause? Because this device was also the one that provided compression for the content going out to customers.

A Layer 7 firewall happily rewrites HTTP headers to reflect the nature of the compressed content (content-length or transfer-encoding: chunked) and to add the gzip flag (accept-encoding:gzip). Since this device was already doing this, it was pretty clear to us that it also had a rule that disabled HTTP persistence for anything with MSIE in the user-agent string.

This was a fine example of the complexity of the modern Web application infrastructure. In effect, there were two groups with different ideas of how Internet Explorer should be handled at the network layer, and neither of them seems to have talked to the other.

When you have a Web performance problem, indulge in a thought experiment. Create an imaginary incoming Web request and try to see if you can follow it through all the systems it touches on your system. Put it on a whiteboard, a mindmap, whatever works.

Then invite the system architects and network engineers in and get them to fill in the gaps.

No doubt that will lead to the “ah ha!” moment. If nothing else, it’s a good excuse to put pizza on the company card. But I have no doubt that you will walk away with a better understanding of your systems, which will make it easier for you to talk to all the people responsible for keeping your systems running.

TAKEAWAY: Just because the part of the Web application you work on is working fine, it may be affected by other components that are not tuned or configured for performance. Get to know the entire application at a high level.

This paper is an extension of the work done for another article that highlighted the performance benefits of retrieving uncompressed and compressed objects directly from the origin server. I wanted to add a proxy server into the stream and determine if proxy servers helped improve the performance of object downloads, and by how much.

Using the same series of objects in the original compression article[1], the CURL tests were re-run 3 times:

1. Directly from the origin server
2. Through the proxy server, to load the files into cache
3. Through the proxy server, to avoid retrieving files from the origin.[2]

This series of three tests was repeated twice: once for the uncompressed files, and then for the compressed objects.[3]

As can be seen clearly in the plots below, compression caused web page download times to improve greatly, when the objects were retrieved from the source. However, the performance difference between compressed and uncompressed data all but disappears when retrieving objects from a proxy server on a corporate LAN.

Instead of the linear growth between object size and download time seen in both of the retrieval tests that used the origin server (Source and Proxy Load data), the Proxy Draw data clearly shows the benefits that accrue when a proxy server is added to a network to assist with serving HTTP traffic.

	MEAN DOWNLOAD TIME
Uncompressed Pages
Total Time Uncompressed — No Proxy	0.256
Total Time Uncompressed — Proxy Load	0.254
Total Time Uncompressed — Proxy Draw	0.110
Compressed Pages
Total Time Compressed — No Proxy	0.181
Total Time Compressed — Proxy Load	0.140
Total Time Compressed — Proxy Draw	0.104

The data above shows just how much of an improvement is gained by adding a local proxy server, explicit caching descriptions and compression can add to a Web site. For sites that do force a great of requests to be returned directly to the origin server, compression will be of great help in reducing bandwidth costs and improving performance. However, by allowing pages to be cached in local proxy servers, the difference between compressed and uncompressed pages vanishes.

Conclusion

Compression is a very good start when attempting to optimize performance. The addition of explicit caching messages in server responses which allow proxy servers to serve cached data to clients on remote local LANs can improve performance to even a greater extent than compression can. These two should be used together to improve the overall performance of Web sites.

[1]The test set was made up of the 1952 HTML files located in the top directory of the Linux Documentation Project HTML archive.

[2]All of the pages in these tests announced the following server response header indicating its cacheability:

Cache-Control: max-age=3600

[3]A note on the compressed files: all compression was performed dynamically by mod_gzip for Apache/1.3.27.

How much improvement can you see with compression? The difference in measured download times on a very lightly loaded server indicates that the time to download the Base Page (the initial HTML file) improved by between 1.3 and 1.6 seconds across a very slow connection when compression was used.

Base Page Performance

There is a slightly slower time for the server to respond to a client requesting a compressed page. Measurements show that the median response time for the server averaged 0.23 seconds for the uncompressed page and 0.27 seconds for the compressed page. However, most Web server administrators should be willing to accept a 0.04 increase in response time to achieve a 1.5 second improvement in file transfer time.

First Byte Performance

Web pages are not completely HTML. How do improved HTML (and CSS) download times affect overall performance? The graph below shows that overall download times for the test page were 1 to 1.5 seconds better when the HTML files were compressed.

Total Page Performance

To further emphasize the value of compression, I ran a test on a Web server to see what the average compression ratio would be when requesting a very large number of files. As well, I wanted to determine what the affect on server response time would be when requesting large numbers of compressed files simultaneously. There were 1952 HTML files in the test directory and I checked the results using CURL across my local LAN.[1]

Large sample of File Requests (1952 HTML Files)

mod_gzip

mod_deflate[2]

Average Compression by File Size mod_gzip mod_deflate 0-999 0.713 0.777[3] 1000-4999 0.440 0.440 5000-9999 0.389 0.389 10000-19999 0.369 0.369 20000-49999 0.350 0.350 50000 and up 0.329 0.331

Conclusion

With a few simple commands, and a little bit of configuration, an Apache Web server can be configured to deliver a large amount of content in a compressed format. These benefits are not simply limited to static pages; dynamic pages generated by PHP and other dynamic content generators can be compressed by using the Apache compression modules. When added other performance tuning mechanisms and appropriate server-side caching rules, these modules can substantially reduce the bandwidth for a very low cost.

[1] The files were the top level HTML files from the Linux Documentation Project. They were installed on an Apache 1.3.27 server running mod_gzip and an Apache 2.0.44 server using mod_deflate. Minimum file size was 80 bytes and maximum file size was 99419 bytes.

[2] mod_deflate for Apache/2.0.44 and earlier comes with the compression ratio set for Best Speed, not Best Compression. This configuration can be modified using the tips found here; and starting with Apache/2.0.45, there will be a configuration directive that will allow admins to configure the compression ratio that they want.

In this example, the compression ratio was set to Level 6.

[3] mod_deflate does not have a lower bound for file size, so it attempts to compress files that are too small to benefit from compression. This results in files smaller than approximately 120 bytes becoming larger when processed by mod_deflate.

The last time I attempted to compile mod_gzip into Apache, I found that the instructions for doing so were not documented clearly on the project page. After a couple of failed attempts, I finally found the instructions buried at the end of the ChangeLog document.

I present the instructions here to preserve your sanity.

Before you can actually get mod_gzip to work, you have to uncomment it in the httpd.conf file module list (Apache 1.3.x) or add it to the module list (Apache 2.0.x).

Now there are two ways to build mod_gzip: statically compiled into Apache and a DSO-File for mod_so. If you want to compile it statically into Apache, just copy the source to Apache src/modules directory and there into a subdirectory named ‘gzip’. You can activate it via a parameter of the configure script.

 ./configure --activate-module=src/modules/gzip/mod_gzip.a
 make
 make install

This will build a new Apache with mod_gzip statically built in.

The DSO-Version is much easier to build.

 make APXS=/path/to/apxs
 make install APXS=/path/to/apxs
 /path/to/apachectl graceful

The apxs script is normally located inside the bin directory of Apache.

NOTE: This hack is only relevant to Apache 2.0.44 or lower. Starting with Apache 2.0.45, the server contains the DeflateCompressionLevel directive, which allows for user-configured compression levels in the httpd.conf file.

One of the complaints leveled against mod_deflate for Apache 2.0.44 and below has been the lower compression ratio that it produces when compared to mod_gzip for Apache 1.3.x and 2.0.x. This issue has been traced to a decision made by the author of mod_deflate to focus on fast compression versus maximum compression.

In discussions with the author of mod_deflate and the maintainer of mod_gzip, the location of the issue was quickly found. The level of compression can be easily modified by changing the ZLIB compression setting in mod_deflate.c from Z_BEST_SPEED (equivalent to “zip -1″) to Z_BEST_COMPRESSION (equivalent to “zip -9″). These defaults can also be replaced with a numeric value between 1 and 9. A “hacked” version of the mod_deflate.c code is available here. In this file, the compression level has been set to 6, which is regarded as a good balance between speed and compression (and also happens to be ZLIB’s default ratio). Some other variations are highlighted below.

zRC = deflateInit2(&ctx->stream, Z_BEST_SPEED, Z_DEFLATED, c->windowSize, c->memlevel, Z_DEFAULT_STRATEGY);

1. zRC = deflateInit2(&ctx->stream, Z_BEST_COMPRESSION, Z_DEFLATED, c->windowSize, c->memlevel, Z_DEFAULT_STRATEGY);

2. zRC = deflateInit2(&ctx->stream, 6, Z_DEFLATED, c->windowSize, c->memlevel, Z_DEFAULT_STRATEGY);

3. zRC = deflateInit2(&ctx->stream, 9, Z_DEFLATED, c->windowSize, c->memlevel, Z_DEFAULT_STRATEGY);

In a previous paper, the use of mod_gzip to dynamically compress the output from an Apache server. With the growing use of the Apache 2.0.x family of Web servers, the question arises of how to perform a similar GZIP-encoding function within this server. The developers of the Apache 2.0.x servers have included a module in the codebase for the server to perform just this task.

mod_deflate is included in the Apache 2.0.x source package, and compiling it in is a simple matter of adding it to the configure command.

	./configure --enable-modules=all --enable-mods-shared=all --enable-deflate

When the server is made and installed, the GZIP-encoding of documents can be enabled in one of two ways: explicit exclusion of files by extension; or by explcit inclusion of files by MIME type. These methods are specified in the httpd.conf file.

Explicit Exclusion

SetOutputFilter DEFLATE
DeflateFilterNote ratio
SetEnvIfNoCase Request_URI .(?:gif|jpe?g|png)$ no-gzip dont-vary
SetEnvIfNoCase Request_URI .(?:exe|t?gz|zip|bz2|sit|rar)$ no-gzip dont-vary
SetEnvIfNoCase Request_URI .pdf$ no-gzip dont-vary

Explicit Inclusion

DeflateFilterNote ratio
AddOutputFilterByType DEFLATE text/*
AddOutputFilterByType DEFLATE application/ms* application/vnd* application/postscript

If a client sends a request which does not include the Accept-Encoding: gzip header, then the item which is stored in the cache cannot be returned to the requesting client if the Accept-Encoding headers do not match. The request must then be passed directly to the origin server to obtain a non-encoded version. In effect, proxy servers may store 2 or more copies of the same file, depending on the client request conditions which cause the server response to Vary.

Removing the Vary response requirement for objects not handled means that if the objects do not vary due to any other directives on the server (browser type, for example), then the cached object can be served up without any additional requests until the Time-To-Live (TTL) of the cached object has expired.

In examining the performance of mod_deflate against mod_gzip, the one item that distinguished mod_deflate from mod_gzip in versions of Apache prior to 2.0.45 was the amount of compression that occurred. The examples below demonstrate that the compression algorithm for mod_gzip produces between 4-6% more compression than mod_deflate for the same file.[1]

Table 1 — /compress/homepage2.html

Table 2 — /documents/spierzchala-resume.ps

Attempts to increase the compression ratio of mod_deflate in Apache 2.044 and lower using the directives provided for this module produced no further decrease in transferred file size. A comment from one of the authors of the mod_deflate module stated that the module was written specifically to ensure that server performance was not degraded by using this compression method. The module was, by default, performing the fastest compression possible, rather than a mid-range compromise between speed and final file size.

Starting with Apache 2.0.45, the compression level of mod_deflate is configurable using the DeflateCompressionLevel directive. This directive accepts values between 1 (fastest compression speed; lowest compression ratio) and 9 (slowest compression speed; highest compression ratio), with the default value being 6. This simple change makes the compression in mod_deflate comparable to mod_gzip out of the box.

Using mod_deflate for Apache 2.0.x is a quick and effective way to decrease the size of the files that are sent to clients. Anything that can produce between 50% and 80% in bandwidth savings with so little effort should definitely be considered for any and all Apache 2.0.x deployments wishing to use the default Apache codebase.

[1] A note on the compression in mod_deflate for Apache 2.044 and lower: The level of compression can be modified by changing the ZLIB compression setting in mod_deflate.c from Z_BEST_SPEED (equivalent to “gzip -1″) to Z_BEST_COMPRESSION (equivalent to “gzip -9″). These defaults can also be replaced with a numeric value between 1 and 9.

More info on hacking mod_deflate for Apache 2.0.44 and lower can be found here.

I have been running the GrabPERF Compression and Performance study for less than a week, but I thought that I should share some of the initial results with everyone.

As you can see above, the byte transmission savings gained by some sites is pretty astounding. Google News sends a pages with a median weight of near 31,000 bytes when compressed; but when compression is disabled on the client, this jumps to over 139,000 bytes.

What is interesting is that the performance gains don’t look truly significant. However, they compressed pages are faster, and have the added benefit of costing the site less, as bandwidth costs count by the byte (I know it’s more complicated than that, but for now, let’s assume a fantasy world).

I will continue to monitor that results and will close the measurements after 14 days and write up a final report.

Technorati Tags: GrabPERF, Web+performance, Web+Compression, HTTP+Compression, Performance+tuning, GZIP

Ok, the Blogger blogs all use GZIP compression.

BUT! The Blogger.com homepage does not.

Ummm…attention to detail? Anyone?

Technorati Tags: Blogger, Upgrade, HTTP+Compression, Web+compression, QA, GrabPERF, Web+Performance

A few years ago, I wrote an article on how GZIP compression improved Web performance. Don Marti at the Linux Journal was a great editor, and eventually, the article ended up in the online version of the Magazine.

At the time, I used Ian Holsman’s webperf.org (now renamed ITScales) to capture the data. Now that I have built my own Web performance monitoring network, I thought I would repeat the experiment.

You can see the comparative results at these locations:

• Wikipedia
• Google News
• Technorati
• Juniper Networks — they own the Redline hardware acceleration device
• F5 Networks — they sell the F5 hardware acceleration device

After I have collected a lot more data, I will be re-visiting the article and commenting on the state of compression technology on the Internet.

If you would like to suggest a site to measure, please leave a comment.

Technorati Tags: HTTP, Web+Compression, Web+performance, GrabPERF, study, HTTP+compression

Dear PodTech:

Say thank you.

HTTP/1.1 302 Found
Date: Tue, 13 Jun 2006 19:18:11 GMT
Server: Apache/2.0.53 (Fedora)
Location: http://www.podtech.net
Content-Length: 283
Content-Type: text/html; charset=iso-8859-1

HTTP/1.1 200 OK
Server: Apache/2.0.53 (Fedora)
X-Powered-By: PHP/4.3.11
X-Pingback: http://www.podtech.net/xmlrpc.php
Status: 200 OK
Content-Type: text/html; charset=UTF-8
Vary: Accept-Encoding
Content-Encoding: gzip
Content-Length: 9233
Date: Tue, 13 Jun 2006 19:14:21 GMT
Connection: keep-alive

Performance got better, didn’t it?

Technorati Tags: GrabPERF, Web performance, PodTech