| Project Status: Active (v0.6.1) | Last modified: April 23, 2002 |
Project Description
|
Abstract: The Zomb WebDev Benchmark (pronounced "Zoom", as in Tomb) is designed to compare the real-world performance of different Web Development languages. A subset of the results can also be used to show the performance delta between different OSs and hardware configurations as well.
|
In its current alpha state, only two languages are incorporated into The Zomb Web Development Benchmark: Perl and PHP. While these languages are cross-platform, the scripts have only been tested on Windows 2000 Professional (v5.00.2195 SP2) using IIS5. Testing will begin on Linux after the bulk of the tests have been completed. Development will continue with only these two languages until a firm benchmarking scheme has been created; doing so allows the project to take place in a less complicated environment. As it stands, ASP/VBScript, C++ and JSP will also be incorporated into zbm as well.
Once a complete version of the benchmark is available, it will be released to the public. This version will include the ability to upload the results to our server so the data can be added to our database; the more results we recieve, the more accurate and interesting the extrapolated finding will be.
Now for a little more about the benchmark itself. We use the term Chapter to describe a particular test scheme. This scheme is then represented in each language that that the benchmark is testing. For example, Chapter 1 is called "Hello, World" and, as you could guess, the scheme for this test is to simply print the line "Hello, World" to the screen; each language has a different syntax, but the scheme or, result is the same.
There are many different ways to approach the problem of how to benchmark, especially when multiple languages are being tested. Should we code the identical functionality in each language? or should we use the tools and shorcuts available in each language to maximize the efficiency? We chose the second approach. Why? Well, we decided that since we are setting out to test the strengths and weaknesses of each language in "Real-World" situations, it makes sense to maximize the potential for each language. Because of this, you will find that the code for each language in a particular chapter can look very different. This is an intentional and conscious decision on our part.
One thing that begs to be mentioned is that the decisions we have made about the benchmark and its methodology are not set in stone. We have set out to find a solid and reliable solution for both benchmarking a web/database server as well as the language(s) empowering it and as such, we are very open to critism, tips, etc... In fact, we greatly appreciate and encourage this typ eof interaction.
|
Chapter Descriptions
Chapter 1: Hello, World (source: Perl, PHP )
This is the first, and most simple, of all the tests in the suite that tests the start up speed of each language (as inspired by those created by Matt Sergeant).
Chapter 2: DB inserting and updating (source: Perl, PHP )
Chapters 2 and 3 make use of MySQL. This Chapter -- Chapter 2 -- creates and fills a MySQL table:
- receives 100 name/value pairs via POST
- parses the data and stores it in an associative array
- connects to MySQL using the DBI/DBD::mysql modules
- creates a table with 4 fields: id, name, number, and date
- inserts each name/value pair into the table (where name=name, number=value and id and date are generated by MySQL)
- updates each record by adding 1 to each number
Chapter 3: DB Selecting and file writing (source: Perl, PHP )
Chapter 3 extracts data from the MySQL table and uses it to create several files:
- Creates a data file called data_LANG.txt for storing the name/value pairs (the name/number fields)
- Creates a template file called
template_LANG.html that will be used by Chapter 4
- Iterates through each record in the table and adds a line to each file as necessary
Chapter 4: File reading and variable interpolation (source: Perl, PHP )
Chapter 4 uses the data in data_LANG.txt to replace the variable fields in template_LANG.html:
- Opens the data file and reads the name/value pairs into a hash
- Opens the template file and reads the contents into a scalar variable
- Runs a regular expression on the variable to replace the variable fields with the contents of the name/value hash (ie, in perl: $hash{VARIABLE_FIELD})
- Creates a new template file, called
new_template_LANG.html, and saves the modified contents of the old template
Chapter 5: Stressing Arrays of text (100 elements) (source: Perl, PHP )
Chapters 5 - 10 are closely related in that they all test the speed of arrays. Chapter 5 gets down and dirty with a small array of text data:
- a
for loop is used to create a 100 element array -- each element contains a string of 12 random characters
- the array is sorted and then reversed
- an element is removed from the front of the array and inserted to the end of it 100 times (useless circular shifting)
- a new array is created in a
while loop by poping and element from the original array and inserting it into the new one.
- the unique elements in the new array are then tallied
Chapter 6: Stressing Arrays of text (1000 elements) (source: Perl, PHP )
Chapter 6 is identical to Chapter 5 with the exception of the array size (this chapter creates an array of 1000 elements).
Chapter 7: Stressing Arrays of numbers (100 elements) (source: Perl, PHP )
Chapter 7 is identical to Chapter 5 with the exception of the array's contents (this chapter fills the array with numeric data).
Chapter 8: Stressing Arrays of numbers (1000 elements) (source: Perl, PHP )
Chapter 8 is identical to Chapter 5 with the exception of the array's contents and size (this chapter fills a 1000 element array with numeric data).
|
|
