Weekly Edition
Return to the Press page
| |||||||||||||
Server Monitoring With BixData (HowtoForge)
HowtoForge looks at server
monitoring with BixData. "BixData is a system, application, and
network monitoring tool which allows you to easily monitor nearly every
aspect of your servers. It can be used for general reporting, for sending
notifications when problems arise, or for automatic maintenance and repairs
- by executing scripts when errors or particular conditions arise."
(Log in to post comments)
Server Monitoring With BixData (HowtoForge) Posted Jun 20, 2006 20:13 UTC (Tue) by jwb (guest, #15467) [Link] It looks better than nagios, but the authors must be high to think that I'm going to deploy this binary blob, which listens for connections on a TCP port, and includes its own parallel authentication scheme, in my precious datacenter.
People trying to break into the software management business should keep some things in mind. No administrator wants to add YET ANOTHER unmanageable, probably-broken, probably-insecure authentication mechanism onto their network. Long-time administrators prefer software which integrates with their routine software and security update procedures, meaning Debian packages instead of blobs in tarballs.
Lastly I will note that the BixData license prohibits me from ever using the software, because I downloaded the server onto one host, but I'd like to run it on another. According to the License.txt, I am "prohibited from (i) making any other copies of the software" so I guess BixData won't have me as a client.
Forget 'em anyway, I say. Nagios seems fine by comparison.
Server Monitoring With BixData (HowtoForge) Posted Jun 22, 2006 15:39 UTC (Thu) by kobus (guest, #38589) [Link] I work on BixData and I appreciate the comment.
BixData is flexible in that you dont have to use the agents. BixServer can monitor a number of services remotely (http, pop, smtp, ports) without requiring an agent to be installed.
Some environments require closer monitoring. If you have an environment with lets say 200 machines running a webcrawler, and would like to be notified when your webcrawler process dies, its a good idea to run something on the machine that will notify you, or a central authority, when the webcrawler process dies, instead of polling it remotely from 1 single location.
Another example is where you'd like to know how much memory, cpu and disk IO your webcrawler process uses every 5 minutes on each one of these machines. Maybe you'd like a graph or report on it every hour. Remotely polling each machine every 5 minutes for these values (if its at all possible to get these values by not installing something), would create a lot of traffic on the network.
Thats where BixAgent is usually used, to monitor local data points you would not be able to otherwise monitor, and to store performance data, that BixServer can periodically collect in an optimized way, and store in SQL and graph.
We wanted to be flexible, so we support mixed environments using agents or no agents.
Security is a big concern. We use the latest OpenSSL library and you can configure BixAgent to only accept SSL connections.
The agent doesnt require any special permissions or changes outside of its install directory, so its quite simple to make a Debian or other install package.
I'm not a legal expert, but the license states that you can not run multiple copies of the software on a single machine.
Bix? (Somewhat OT) Posted Jun 20, 2006 21:22 UTC (Tue) by AJWM (guest, #15888) [Link] I can't help but be reminded of the old Byte Information eXchange started about 20 years ago by Byte Magazine -- a text-based conferencing (mostly) system that eventually gave up competing with the web about 6 or 7 years ago.
Their web site is unhelpful as to where the name came from.
(The CoSy conferencing system code that BIX was based on was GPL'd a few years ago, avalailable on Sourceforge. Some of the die-hard bixen used it to start NLZero (nlzero.com)).
As for monitoring, I use a mix of OV (HP OpenView) and nagios.
Interesting but.. Posted Jun 20, 2006 21:48 UTC (Tue) by yem (guest, #1138) [Link] There are a few projects which seek to merge monitoring of performance (a la cacti) and avalability (a la nagios) into one glorious application and they usually fall short. Nagios is hairy, I'll admit, but it has a modular config system, simple extensibilty (we have a tonne of custom tailored check and alert scriptlets) and never breaks down. It'll take more than a native GUI to replace nagios.
Interesting but.. Posted Jun 20, 2006 23:12 UTC (Tue) by gbailey (guest, #58) [Link] Agreed. However, I did recently try OpenNMS and found it had matured a LOT since I had last looked at it 3-4 years ago. Worth a look for those seeking a performance/monitoring combination...
Server Monitoring With BixData (HowtoForge) Posted Jun 21, 2006 18:09 UTC (Wed) by NightMonkey (subscriber, #23051) [Link] Ganglia (http://ganglia.sourceforge.net/) is an incredible monitoring system. It's what Wikipedia uses, among hundereds of other large sites. And no "binary blobs".
Server Monitoring With BixData (HowtoForge) Posted Jun 22, 2006 5:48 UTC (Thu) by _pupu_ (guest, #38572) [Link] We're using Nagios for remote monitoring (services' test from behind a firewall/content switch, SNMP load check etc) and Monit (http://www.tildeslash.com/monit) for local monitoring. I will probably try BixData, but Monit does everything we need (monitoring services, executing scripts, sending alerts) with nice and easy configuration. Plus it is GPLed with a possibility of commercial support. Well, we'll see.
|
|||||||||||||