Compression Tools Compared
Data compression also can speed up network transfers. How much depends on how fast your CPU and network are. Slow networks with fast CPUs can be sped up the most by thoroughly compressing the data. Alternatively, slow CPUs with fast connections do best with no compression.
Find the best compressor and compression level for your hardware in the graph shown in Figure 6. This graph's CPU and network speed axes are scaled logarithmically too. Look where your CPU and network speeds intersect in the graph, and try the data compression tool and compression level at that point. It also should give you a sense of how much your bandwidth may increase.
Network Transfer Estimates
To find the best compressors for various CPU and network speeds, I considered how long it takes to compress data, send it and decompress it. I projected how long compression and decompression should take on computers of various speeds by simply scaling actual test results from my 1.7GHz CPU. For example, a 3.4GHz CPU should compress data about twice as fast. Likewise, I estimated transfer times by dividing the size of the compressed data by the network's real speed.
The overall transfer time for non-filtering data compression tools, such as rzip, simply should be about the sum of the estimated times to compress, send and decompress the data.
However, compressors that can act as filters, such as gzip, have an advantage. They simultaneously can compress, transfer and decompress. I assumed their overall transfer times are dominated by the slowest of the three steps. I verified some estimates by timing real transfers.
For example, if you have a 56Kbps dial-up modem and a 3GHz CPU, their speeds intersect in the light-yellow region labeled lzma 26 at the top of the graph. This corresponds to using lzma with a 226 size dictionary. The graph predicts a 430% increase in effective bandwidth.
On the other hand, if you have a 1GHz network, but only a 100MHz CPU, it should be faster simply to send the raw uncompressed data. This is depicted in the flat black region at the bottom of the graph.
Don't assume that you always should increase performance the most by using lzma, however. The best compression tool for data transfers depends on the ratio of your particular CPU's speed to your particular network's speed.
If the sending and receiving computers have different CPU speeds, try looking up the sending computer's speed in the graph. Compression can be much more CPU-intensive. Check whether the data compression tool and scp are installed on both computers. Remember to replace user@box.com and file with the real names.
For the fastest CPUs and/or slowest network connections that fall in the graph's light-yellow region, speed up your network transfers like this:
$ cat file \ | lzma -x -s26 \ | ssh user@box.com "lzma -d > file"
ssh stands for secure shell. It's a safe way to execute commands on remote computers. This may speed up your network transfer by more than 400%.
For fast CPUs and/or slow networks that fall into the graph's dark-yellow zone, use rzip with a compression level of one. Because rzip doesn't work as a filter, you need temporary space for the compressed file on the originating box:
$ rzip -1 -k file $ scp file.rz user@box.com: $ ssh user@box.com "rzip -d file.rz"
The -1 tells rzip to use compression level 1, and the -k tells it to keep its input file. Remember to use a : at the end of the scp command.
rzipped network transfers can be 375% faster. That one-hour transfer might finish in only 16 minutes!
For slightly slower CPUs and/or faster networks that fall in the graph's orange region, try using gzip with compression level 1. Here's how:
$ gzip -1c file | ssh user@box.com "gzip -d > file"
It might double your effective bandwidth. -1c tells gzip to use compression level 1 and write to standard output, and -d tells it to decompress.
For fast network connections and slow CPUs falling in the graph's blue region, quickly compress a little with lzop at compression level 1:
$ lzop -1c file | ssh user@box.com "lzop -d > file"
Realizing the promise of Apache® Hadoop® requires the effective deployment of compute, memory, storage and networking to achieve optimal results. With its flexibility and multitude of options, it is easy to over or under provision the server infrastructure, resulting in poor performance and high TCO. Join us for an in depth, technical discussion with industry experts from leading Hadoop and server companies who will provide insights into the key considerations for designing and deploying an optimal Hadoop cluster.
Sponsored by AMD
Built-in forensics, incident response, and security with Red Hat Enterprise Linux 6
Every security policy provides guidance and requirements for ensuring adequate protection of information and data, as well as high-level technical and administrative security requirements for a system in a given environment. Traditionally, providing security for a system focuses on the confidentiality of the information on it. However, protecting the data integrity and system and data availability is just as important. For example, when processing United States intelligence information, there are three attributes that require protection: confidentiality, integrity, and availability.
Learn more about catching the bad guy in this free white paper.
Sponsored by DLT Solutions
| Designing Electronics with Linux | May 22, 2013 |
| Dynamic DNS—an Object Lesson in Problem Solving | May 21, 2013 |
| Using Salt Stack and Vagrant for Drupal Development | May 20, 2013 |
| Making Linux and Android Get Along (It's Not as Hard as It Sounds) | May 16, 2013 |
| Drupal Is a Framework: Why Everyone Needs to Understand This | May 15, 2013 |
| Home, My Backup Data Center | May 13, 2013 |
- Designing Electronics with Linux
- Making Linux and Android Get Along (It's Not as Hard as It Sounds)
- Dynamic DNS—an Object Lesson in Problem Solving
- Using Salt Stack and Vagrant for Drupal Development
- Build a Skype Server for Your Home Phone System
- Why Python?
- New Products
- A Topic for Discussion - Open Source Feature-Richness?
- Validate an E-Mail Address with PHP, the Right Way
- Tech Tip: Really Simple HTTP Server with Python
- Understanding the Linux Kernel
1 hour 40 min ago - General
4 hours 9 min ago - Kernel Problem
14 hours 12 min ago - BASH script to log IPs on public web server
18 hours 39 min ago - DynDNS
22 hours 15 min ago - Reply to comment | Linux Journal
22 hours 47 min ago - All the articles you talked
1 day 1 hour ago - All the articles you talked
1 day 1 hour ago - All the articles you talked
1 day 1 hour ago - myip
1 day 5 hours ago
Enter to Win an Adafruit Pi Cobbler Breakout Kit for Raspberry Pi

It's Raspberry Pi month at Linux Journal. Each week in May, Adafruit will be giving away a Pi-related prize to a lucky, randomly drawn LJ reader. Winners will be announced weekly.
Fill out the fields below to enter to win this week's prize-- a Pi Cobbler Breakout Kit for Raspberry Pi.
Congratulations to our winners so far:
- 5-8-13, Pi Starter Pack: Jack Davis
- 5-15-13, Pi Model B 512MB RAM: Patrick Dunn
- 5-21-13, Prototyping Pi Plate Kit: Philip Kirby
- Next winner announced on 5-27-13!
Free Webinar: Hadoop
How to Build an Optimal Hadoop Cluster to Store and Maintain Unlimited Amounts of Data Using Microservers
Realizing the promise of Apache® Hadoop® requires the effective deployment of compute, memory, storage and networking to achieve optimal results. With its flexibility and multitude of options, it is easy to over or under provision the server infrastructure, resulting in poor performance and high TCO. Join us for an in depth, technical discussion with industry experts from leading Hadoop and server companies who will provide insights into the key considerations for designing and deploying an optimal Hadoop cluster.
Some of key questions to be discussed are:
- What is the “typical” Hadoop cluster and what should be installed on the different machine types?
- Why should you consider the typical workload patterns when making your hardware decisions?
- Are all microservers created equal for Hadoop deployments?
- How do I plan for expansion if I require more compute, memory, storage or networking?





Comments
How to improve backups?
Hi. Congratulations for this very useful article... I use a script for backup made by myself which use tar +gzip... switching to tar - lsop backup time takes less than half time, increasing the backup size by about 25%.
An idea to improve speed is to replace tar with a more intelligent tool.
Infact, tar simply "cat all files to stdout" and then gzip or lsop compress this huge stream of data, but some data is already compressed (images, movies, open document files) and don't need to be recompressed!
The idea is to have an archiver (like tar) which compress each file by itself, storing the original file in case of images, movies, archives, already compressed files.
Is there any tool that can do this, and save all priviledges (owner, group, mode) associated to each file like tar does?
Thank you. Paolo
(1) Found a typo: "On the
(1) Found a typo:
"On the other hand, if you have a 1GHz network, but only a 100MHz CPU"
1 GHz network? Should maybe be 1 Gbps.
(2) Suggestion:
Multi-Core CPUs are the big thing today, compression tools that could utilise multiple cores can run 2, 4 or soon even 8 times faster on "normal" desktop PCs...not even speaking of the servers...which compression tools can utilise this CPU power?
multi-core CPU support
Multi-Core CPUs are the big thing today, compression tools that could utilise multiple cores can run 2, 4 or soon even 8 times faster on "normal" desktop PCs...not even speaking of the servers...which compression tools can utilise this CPU power?
http://compression.ca/pbzip2/
There's parallel bzip2, very good but not pipe support.
HTH,
mfg zmi
Very nice information
Very nice information provided.Thanks!!!
Excelent article.
Thanks very much for this article. I really enjoyed it, and will be helpfull for my daily work.
1. how about another part
how about another part with specific data - like 90+% text? for mysql dumps & dbmail scenarios etc.
and 45MB does not sound as sufficient test data size for rzip to test it's speed.
Compression on Windows
First of all excellent test!.
Believe it or not, but compression is one of those application types where all research takes place on Windows Pc's. The last couple of years there were some major breakthroughs in compression caused by the new PAQ context modeling algorithms. Have a look at this site for some results. Programs like gzip, rzip 7-zip and lzop are tested here too, so it should be easy to compare results.
http://www.maximumcompression.com/