Slashdot Log In
Suggestions For Cheap Metrics Eye Candy Software?
Posted by
timothy
on Wed Nov 26, 2008 07:10 PM
from the compiz-plus-millions-of-stats-widgets dept.
from the compiz-plus-millions-of-stats-widgets dept.
Banquo writes "I have a friend who has a small datacenter (SQL/Mail/IIS/File Repository ... 5 or 10 servers) and he was saying that his boss wants to see some kind of 'visual display of changing metrics' — Net/server/sql stats with moving lines and graphs and pretty colors. Basically they want something to display on a big LCD panel that will give a tiny bit of 'Wow' factor to customer visits. Back in my datacenter days I saw a million packages to do this stuff, but I was always blessed with an IT budget for metrics/monitoring. Can anyone suggest a free/cheap package that will make pretty moving pictures, moving lines, graphs, etc. from server/net stats? There's no worry about actually using this for real data tracking or metrics purposes. He has a pretty robust log/alert/metrics setup, but command line is a little too dry for marketing purposes. I jokingly suggested he just use a looped flash animation but he actually does want stats that are coming from and reflect his environment. Anyone know of any cheap or free data center stats/metrics 'Eye Candy' software out there?" Better yet, can you think of any particularly interesting ways to display that sort of information?
Related Stories
[+]
Technology: SGI's Open Source Performance Co-Pilot 24 comments
codesmythe writes "The Fates, through SGI nee Rackable, have granted a new beginning to Silicon Valley's once darling Silicon Graphics. Despite old mistakes and economic misfortunes, Silicon Graphics' engineering contributions are legendary: their systems (oh, the systems!), and software such as the well known OpenGL and the little known Performance Co-Pilot. PCP is an enterprise-class open source system monitoring, measurement, and visualization infrastructure — overlooked in last fall's monitoring tool discussion. Since its proprietary beginning in 1993, PCP has been re-released as open source and ported to all major operating systems. Readers of Slashdot's recent Beginning Python Visualization book review will be pleased to hear there are Python interfaces to PCP data sources. Here is an example of using Python and Blender to visualize PCP data (registration may be required). The PCP dev community is well and active, and includes several of the original team members."
This discussion has been archived.
No new comments can be posted.
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
Full
Abbreviated
Hidden
Loading... please wait.
rrdtool. (Score:5, Informative)
Pandora FMS (Score:5, Informative)
One option I'm reviewing at the moment is Pandora FMS
http://pandora.sourceforge.net/ [sourceforge.net]
Not bad and there's a pre-built vm you can download to quickly give it a go.
http://www.vmware.com/appliances/directory/1236 [vmware.com]
Parent
Re: (Score:3, Informative)
and maybe one of the projects that use it [oetiker.ch].
mrtg and rrdtool are the grand daddies of the monitoring eye candy set. http://oss.oetiker.ch/ [oetiker.ch] for all the projects he made.
Re: (Score:3, Funny)
as you can see from this graph [oetiker.ch] our Santa projections for next quarter are very promising.
Re:rrdtool. (Score:5, Interesting)
rrdtool is great to show a graph of disk usage and so forth - for management of servers and for management of managers - but showing it to clients on a tour? Big whoop. Any hick can go make a graph (I personally graph
Anyone have a link to the google projector where they throw up the current search term on the wall? Completely useless but freaking awesome. That's the sort of thing you want to show clients, not a bunch of graphs about bandwidth usage and CPU speed.
Parent
Re: (Score:3, Informative)
Screensaver (Score:5, Funny)
Re:Screensaver (Score:5, Funny)
Parent
Re:Screensaver (Score:4, Funny)
I don't even see the code anymore, all I see is bomb, BSOD, kernel panic...
Parent
Re:Screensaver (Score:4, Interesting)
A friend of mine recently told me they were using lava lamps in different colors as a low-tech indicator for problems with the automatic overnight test/build process. A developer would enter the office in the morning and immediately notice an eerie yellow glow, which meant that the test suite for project #2 didn't complete successfully. He'd know he'll have to look into that even before checking his email (after making some coffee, reading Slashdot and doing the rest of his early morning routine). Might be a bit too geeky for customers, but from what I heard, it works quite well.
Parent
GL Tail (Score:5, Informative)
Re: (Score:3, Informative)
Look at Munin (Score:5, Interesting)
A dozen xterms... (Score:5, Funny)
...each running 'tail -f' on a log file.
Re: (Score:3, Interesting)
You laugh, but we had exactly this installed at my last place and we knew instantly if something was wrong, either by noticing odd patterns in the text or by one stopping completely.
Nisca (Score:5, Interesting)
I found Nisca [sourceforge.net] better and easier to extend than rrdtool. I liked the fact it has full history so you can zoom in on the stats at any point in the past. But it is a difficult to set up for the first time and seems half-abandoned now.
Lies (Score:4, Insightful)
--
Lost your job? Keep one eye open on craigslist, even just for gigs http://www.bigattichouse.com/oneeyeopen.html [bigattichouse.com]
Re:Lies (Score:5, Insightful)
Parent
Re:Lies (Score:4, Insightful)
Parent
Not free but pretty cheap (Score:4, Interesting)
cacti (Score:5, Interesting)
I just grabbed a Cacti virtual appliance from rPath. No installation required really - just load it into VMWare (you can also get isos) and configure it. No chasing down prereqs or dependencies. I'm not affiliated, just impressed with the ease.
http://www.rpath.com/rbuilder/ [rpath.com]
Re:cacti (Score:4, Informative)
Parent
Ganglia (Score:3, Informative)
We use Ganglia (http://ganglia.info) at work.
If you prefer command line, try nmon. Originally for AIX, but there's a Linux port. Works well. On a large green-on-black terminal it looks pretty cool :D
Re: (Score:3, Funny)
Quartz Composer (Score:5, Interesting)
If the LCD panel is connected to a Mac, you may want to try using Quartz Composer.
It's a flow-based programming language included in the developer tools package. You can use it to make just about any kind of animation (music visualizations, image filters, screensavers, etc.), and hook it up to live data.
I've set it up for my office, but didn't have time to write a very complex program yet, just a flashy 3D RSS feed of Twitter posts mentioning our product.
Spotlight on Windows (Score:5, Informative)
Jesus, did any of you even RTFS? I'd hate to see software requirements from any of you fools.
He asked for moving pictures and lines:
Quest's Spotlight on Windows.
Screenshot at http://www.quest.com/images/popup.asp?path=/spotlight_on_windows/img/screenshots/5.png&width=1280&height=993
MRTG (Score:3, Informative)
MRTG can graph pretty much anything. It's primarily used for bandwidth (I think- given the name), but a former company used it to graph pretty much everything about all its servers: CPU load, motherboard temperature, bandwidth, disk capacity, web server hits, mail system access. It's written in perl and pretty easy to customize, from what I understand; essentially, anything that can dump two numbers into a file can be used to produce a graph, and the look and feel of the graph can be changed in the config.
good summary here: (Score:4, Interesting)
Short list (Score:5, Informative)
Ntop [ntop.org].
Nagios [nagios.org].
MRTG [oetiker.ch].
Severed Head of PHB (Score:3, Funny)
The best sort of visual indication of status to the PHB is the severed head of another PHB on a spike at the entrance to the data centre.
webminstats (Score:4, Informative)
Nagios+R2D2 (Score:3, Funny)
I know nothing about Nagios. But whatever you do, it should be displayed via R2 Unit [youtube.com]
Saw it bring a network down once... (Score:5, Interesting)
It was too funny. Some other chief sales drone insisted they wanted pretty dancing graphs like a stereo equalizer, so the cheap-salary french fry maker/network engineer in charge of it turned on every SNMP query possible at the core, dug up the command to give SNMP queries the highest possible priority, and then set their SNMP monitoring tool to query everything about a dozen times a second.
CPU Utilization, which was already at a heavy 70%, pegged. The whole network shuddered to a screaming halt. Trouble tickets flooded in, customers and everyone else screaming bloody murder...
Naturally, Fate saw to it this issue hit my desk. "Why," I asked, rubbing my temples and already fearing the answer, "did you do this?"
"They wanted it to look cool."
I raised me voice loud enough for the room to hear. "I'm sorry, we had some static, I didn't catch that. Could you repeat that?" Everyone fell silent as I hit the "speaker" and then "mute" buttons on my phone.
"I wanted it to look cool, you know, like 'the Matrix?'"
Everyone got a merrily constipated look on their face. One of my buddies across the room asked "We on mute?"
"Of course."
The room full of CCIEs laughed for a good three minutes. For weeks afterward, "I wanted it to look cool, like the Matrix" was a catch phrase.
wow (Score:5, Insightful)
you're a dick. given that this guy is low salary he probably doesn't have a lot of experience. you could have shown him the error of his ways, instead you publicly embarrass him in front of the whole company. glad I don't work with you.
Parent
Re:wow (Score:4, Insightful)
you're a dick. given that this guy is low salary he probably doesn't have a lot of experience. you could have shown him the error of his ways, instead you publicly embarrass him in front of the whole company. glad I don't work with you.
On the one hand, you're right. Embarrassing the idiot was clearly a dick move.
On the other hand, this is a very useful bit of dickishness. The idiot didn't just make a mistake; he made a mistake with major consequences to a lot of people, and he made a mess that his betters had to clean up.
In my experience, about 98% of the time, there are only two ways we learn. One is through pain. The network breaker, among many flaws, had insufficient caution, but I'm sure the pain of humiliation here taught him some. (That's one of the skills he'll need if he ever wants to be a highly paid admin.) The other way is through observing the pain of others. By making a semi-public example of the yutz, a room-full of network engineers (and I'm sure, a lot of their friends) got a great example of how not to behave. You can bet that at least some minor fuckups were avoided thanks to this.
Sysadmins are often dicks to fools for a reason: it helps a lot in their work. I didn't like hating everybody all the time, so now I'm a recovering sysadmin [faqs.org]. Bitch all you want, but however unforgiving sysadmins are, the machines they run are far less so.
Parent
Re:wow (Score:4, Insightful)
This touched a nerve. I've been sysadmin'ing for a long time now (well, not THAT long. 10 years or so), and I've seen my share of abusive system administrators, it annoyes me every single time.
People don't learn anything useful from pain, they only learn behaviorism - and then they learn that their senior system administrators is some elitist assholes. Okay, the latter is somewhat useful to know.
Many system administrators are exactly as unforgiving as the machinery they run - and it don't have to be that way. System administrators must provide (as everybody in IT) vertical support for the entire organization, not the other way around. Many system administrators don't realize this. Instead they only accept one truth. Their own.
Parent
Re: (Score:3, Insightful)
Yes. He was abusive when he deliberately humiliated a coworker for no apparent reason other than having a laugh with the other smugs.
The Matrix (Score:5, Insightful)
Top, and watch tail logfile really impress people.
This will make it look like you/your techs are amazing, and doing things that noone can conceive of. Pie charts and graphs make the job look easy, and noone wants to pay for easy.
My 2 cents.
BSOD network visualisation (Score:5, Informative)
Use Processing (Score:3, Interesting)
Try coding it up in Processing [processing.org]
You could visualize events as swarming butterflies! [vimeo.com]
Use CSS and meta-refresh (Score:3, Interesting)
A List Apart has discussed this at length.
http://www.alistapart.com/articles/accessibledatavisualization/ [alistapart.com]
Generating overlapping squiggly lines is a small variation on the spark charts (you're just placing 1px high objects)
Personally I'm using Tiny webserver and a dozen lines of Perl (yes, I'm old) to provide similar functionality.
For display, play with your IE/Opera/Ffox window toolbar settings to get rid of everything bar the screen and job's done.
In my case, the fun part is getting the data out of Wireshark (http://www.wireshark.org/) automatically :-)
The early version (Score:3, Informative)
Way too many years ago, in 1971, I did something like this for a UNIVAC 1108 mainframe. We had a big CRT hanging from the ceiling of a glass-walled computer room, showing some basic information like current job status, memory utilization bar chart, backlog, and console messages. Every four seconds, the display changed to a new screen.
People would actually come up to the glass wall to watch. For the first time, there was some indication of what the mainframe was doing. The mainframe's console was a teletype, and the operator could make some status inquiries, but at 110 baud, you couldn't get mucn insight into what was going on. (That operating system viewed the operator as a peripheral; most of what appeared on the console consisted of orders for the operator to mount tapes, change paper in printers, and such.)
Today you need more entertainment value. If you want something really cool, you might try outsourcing the job to a Flash developer. Provide some way for Flash to get the needed data, and do all the eye candy in Flash.
Oh, that thing! (Score:3, Informative)
About, oh two years ago, there was a slashdot article about someone who had built something PERFECT! It was open source, and I spoke with the gentleman, who's willing to alter it for you if you haven't the time. It was basically something written for linux, I think it was written in perl with its graphics thing.
It basically had two columns, one on either side of the screen, each being a list of somethings. URLs, recent humans, whatever. And every time a web-page was served, it spit out a little round circle, the size was proportional to the time to generate the page, or the amount of data sent, or whatever. And then certain events, like a user login, or a purchase, appeared as text faded in, and then flew upwards.
The system was designed to work with any data source, not just web stats.
I remember little more. There was a little video showing it in action.
xmms (Score:3, Funny)
True story. We had some clients coming to town for a visit and I was asked to put some fancy monitoring system in the server room. So I hooked a notebook to an external monitor, copied some mp3s onto it, and ran xmms with a bunch of spectral analyser add-ons. It looked very high-tech, and everyone was impressed. Of course I didn't tell them that it was "monitoring" Avril Lavigne music 24/7.
How about a screensaver doing this? (Score:3, Interesting)
If you think about it, quite a few systems have screens you only need when something's gone wrong.
If you have a screensaver on a tech display that picked up the vital statistics from somewhere you would have the display, but also the use of the screen when something blows up with autmotic resumption when you stop working on the system. In principle should the screensaver simply be the remote display (so you could choose what to display where, or even build a collection of stats for one screen). The main disadvantage is, of course that this won't "save" much screen :-), and you may need a permanent copy somewhere that won't vanish when you touch the keyboard..
A good decade ago I had a 30 user PowerLAN setup (yes, ARCnet :-), and the server screen was a simple, ASCII based set of graphics showing server load, network load and disk capacity in log based bars (more sensible than straight linear representations), and other relevant data in numbers. I still think that was one of the most sensible server displays ever but it did a good job of burning in the CRT when we forgot the powersave :-)
Re: (Score:3, Insightful)
It's rather difficult to coordinate several hundred stats at minute resolution and make things move like a speedometer. RRDTool is fantastic, present it the right data and all is good. When there are many pretty widgets to look at, 5 minute resolution is often better than good enough. RRDTool can be used to display aggregated RRD data as well, so you can have simple go/no_go indicators as well as pretty widgets. A bit of PERL and you can do wonderful things with data fed to RRDTool.
logs jgraph (Score:4, Interesting)
Parent
Re:you don't say.. (Score:4, Insightful)
Parent
Re: (Score:3, Informative)
If you want something cool with multidimensional data, do something with GGobi [http://www.ggobi.org/]
Re: (Score:3, Funny)
How about that software that plays music, and is attuned to the load of each server?
You could tie it in with the lighting and environmental controls - if someone walks in on you and the lights are dim, it's hot as hell, and there's dramatic music in the background they know to leave you alone. Perhaps some torture-chamber sound effects could be included.