DUG - Daf's Uncomplicated Grapher

1. What is DUG?
2. Screenshots
3. A demonstration
4. Who is DUG for?
5. How do I use DUG?
6. Cool! How do I get DUG?

Hosted on

What is DUG?

DUG is a snapshot data graphing utility designed to be as simple as possible whilst still being as powerful as required. By "snapshot data", I mean data that DUG can query an external source for on a timed basis to build up snapshots of the data throughout the course of time. DUG is written in C++ and works nicely under WIN32, WIN64 (32bit binaries -- 64bit binaries have not been tested, but should work fine), Linux (32bit and 64bit).

DUG had a much cooler name but unfortunately there's a dormant project with that name. I would really like it if the SourceForge team would have systems in place to remove dead and long-standing dormant projects. There are many empty projects on SourceForge which have been registered and have little or no activity for years. Like IP addresses, the good names are running out. DUG was renamed out of frustration waiting for the project name that I wanted to become available. The short of it is this: I'm Daf, I wrote DUG, it's meant to be easy / simple to use (uncomplicated) and it draws graphs. Exciting stuff.

DUG currently supports SNMP and XML data sources. DUG has only been fully tested with SNMP version 2c. Other versions may or may not work. Version 2c is the most common version to be found "in the wild".

DUG can be used to graph information like:

• Bandwidth usage on a Windows or Linux computer; or on any remote SNMP-capable device or any remote system which can generate an XML report for consumption
• System information like temperatures, disk usage, etc. If the data you seek is not exported via SNMP, you can write your own scripts or applications to present that data in XML form. DUG can consume XML via http, https, ftp, or local files.
• Basically, ANY numeric data which can be queried at a timed interval from an SNMP or XML source can be graphed

DUG can be used instead of MRTG or CACTII (which is built on MRTG). DUG was originally designed to provide similar functionality to that of MRTG with an easier-to-configure interface. DUG expanded beyond that original scope to include XML sources. DUG, by default, has some nice features not present (or difficult to reproduce) with MRTG:

• INI-style files for configuration: easier to read and edit by humans than XML.
• Extensive override system in the configuration to allow the least configuration possible as well as the most control possible.
• Sane defaults for most things -- again, to reduce unnecessary editing
• Stacked graphs show how individual data sources contribute to the whole. See my first screenshot below for an idea
• Small configuration files go a long way. To get 4 graphs (Daily, Weekly, Monthly, Yearly) for incoming and outgoing bandwidth on a network interface via SNMP takes about 10 lines of INI-style configuration. Adding another NIC could be as little as another 2 lines.
• Statistics for the monitored data sets. Not only can you see a graph of bandwidth usage, for example, but you can see the total, average, current (at time of graphing), min and max for the period being graphed. Easily see how much of your monthly bandwidth has been used. Get actual figures to show how trends have changed over time.
• SVG output means graphs that scale. SVG is the default output medium, although DUG can be configured to use an external too, like RSVG, to convert SVG images to PNG images for the one and only browser yet to support a well-established internet standard: Internet Explorer.
• By default, all data is kept for a year, which means that you can generate reports for any timeframe within the last year where DUG was active. MRTG throws your old data away -- DUG allows you to use it again. To save disk space, DUG has options to lighten the database -- but disk space is cheap, and time-walking is cool.
• No dependency on external databases. No need to install MySQL. DUG uses SQLite for the backend.
• Easy install and minimal to no dependencies on external libraries, especially for WIN32. This means that you can deploy dug.exe on your server by simply copying it there. No setup. No extra downloads. Linux servers generally provide better build environments. Building under Linux is quite painless.

Screenshots

Pictures often speak louder than words. Here are two graph sections from my home report:

This shows the bandwidth usage on my home machine. The darker orange is download and the lighter orange is upload. Note that I have quite a constant upload because I throttle my outgoing torrents due to my terrible upload speed -- hence the fairly constant 1kB/s. The graph here is also rendered in stacked mode such that the overall shape of the graph shows the total bandwidth usage (in and out) whilst the individual directions' usage can be inferred from looking at the graph. Note that DUG chose the colors for this graph -- I just left them unconfigured and DUG chose random colors, adding them to the config file

This shows my CPU temperature history on my home machine. I get my CPU temperatures via SNMP.

Note that all of the page except the dataset have taken on my system colors. DUG tries to "fit in" with the user's preferences. That being said, DUG is fully open for a web developer to alter without delving into the source code of DUG itself. DUG will honor a stylesheet with the name "style.css" in the same dir as the report. DUG can also inline an arbitrary CSS file mentioned by path in the report's INI file, or can use per-report css files that are found in the report's output dir, providing the name of the css file is the same as that of the report (eg "foo.css" for report "foo.html").

DUG can also take extra javascript, via similar mechanisms ("foo.js" for report "foo.html") and a dir-wide "dug.js" in the report dir. Once again, this is to assist eager web developers who don't want to meddle with the DUG C++ source.

A demonstration

There's an interactive demo of DUG output here. Bear in mind that DUG generates static output, so the data will quickly become dated. The interactivity is in the page itself -- you can't regenerate the report. Also, I've included a graph which is very young to demonstrate an XML source. The HDD temp graph gets its data from output from this python script which basically just uses the commandline hddtemp tool to build a very simple XML document like this. The INI file used to create this report is here.

Who is DUG for?

The short answer is: anyone who wants to see trends in data which is available via SNMP or XML. This means:

• System / network administrators who need to monitor server data
• Home users who want to monitor how much bandwidth they have consumed over specified time periods. Sadly enough, not all of us can have unlimited bandwidth. The reality is that many people have capped bandwidth with either penalties for going over the capped amount or an abrupt denial of service. DUG provides an easy way to monitor your cap.
• Curious people who have noticed up-to-date numeric data in HTML pages on the internet. DUG's XML parser is very lenient -- DUG will handle HTML4 as if it were "well-formed" XML. DUG's XML generator, on the other hand, is very strict -- DUG output is quite pure XHTML which has passed HTML Tidy tests for HTML, XML and XHTML with no errors or warnings. DUG uses XPATHs to get to information in a document. DUG's implementation of XPATH may not be 100% complete, but works for most use cases. XPATH feature requests will be considered.

How do I use DUG?

DUG tries to almost be a good UNIX tool -- it does two things, and does them fairly well:

1. Capture data
2. Draw graphs and stats

DUG relies on the person setting it up to schedule DUG to be run as required. You would schedule DUG via cron or Task Scheduler to capture and draw as required. This way, data granularity and system resource usage is controlled by you -- not DUG. Capture more often, and your data will be finer, but your database will be larger. Graph generation will take a little longer. Capture less often, and your database is smaller and graph generation is faster but graphs will be "coarser".

Some examples of usage intervals and expected experiences are as follows:

• At a large-ish company where 200 virtual hosts are monitored for outgoing bandwidth consumption, and some total / summary usage information needs to be graphed, DUG is set up to capture data once every 5 minutes and draw once every 5 minutes. The capture here takes about 30 seconds (most of that time is taken waiting for responses from the SNMP sources) and the draw (about 16 reports, about 220 graphs in total) takes around 3 minutes to run. DUG is not extensively threaded, so will only run on one of your cores. After about 6 months of usage, the database is roughly 340mb. SQLite seems to allocate database space in chunks: this number hasn't changed in some time. I estimate that a year's total information will come in at about 500-600 mb for this case. SQLite remains responsive throughout this -- run times seem to have remained fairly constant as that database has grown.
• At home, I use DUG to monitor my internet connection, my dual-core cpu temperatures and my 5 hard drive temperatures. DUG runs once a minute to capture data and once every 5 minutes to draw graphs. The capture takes a second or two and the draw takes about 3 seconds. My home machine is an Athlon64X2 6400+

DUG produces STATIC html and svg (optional conversion to png) output: DUG does not need a web server to be useful. Obviously, if you want to make the output from DUG available over the web, you need a web server. But home users can just point their browsers at the html output from DUG.

DUG was originally a project for work, where I am required to document my work on a MediaWiki site. SourceForge also supplies MediaWiki, so, for your own ease of use, the readme.wiki is also available in formatted form here. The document at this location may also be more up-to-date than its counterpart(s) in packages. If I fix an error, it is most likely at the SF wiki page. My apologies for any discrepencies between the README you have and the one presented here.

How do I get DUG?

DUG is hosted on SourceForge, and can be obtained here. DUG currently does not have a graphical configuration helper -- but one is planned. DUG does not have an official installer either -- simply unzip the contents of the WIN32 release wherever you want it, or build from source.