Software for IT Management
Monday, 11. June 2007, 13:00:47
Now that I am working in IT managing a diverse network of devices, I've aquired some experiance with what might work and might not work across those devices. One thing I've noticed is you can spend a LOT of money on various multi access tools to get some idea of what's going on with your Windows, Linux, and MacOSX machines, plus you might have Solaris, AIX, HP-UX or BSD machines. Fun.
Now, I haven't had the ability to compare various commercial offerings beyond looking at their documentation as each is quite expensive. But for those of you without oodles of dollars, I have identified, and used with some success some FLOSS tools. Some of these are really quite mature, some are usable with great potential. For some reason, many of these seem to come from France.
I could talk for some time on various tools, but for now I'll just lay out what I was looking for, and what I've (well, the team I'm on) have done.
We wanted something that would work across Linux, Windows and MacOSX. More is great, but those were important - we wanted to consolidate our management tools, not increase them. We already have management for each platform separately except for inventory of some sort, so that's were we started - inventory. And there is a great package of two projects that does this nicely, and has a useful package deployment tool as well.
OCSNG + GLPI. The "names" for the two projects.
ocsinventory-ng.org
glpi-project.org
So, you have to have OCSNG - you can layer GLPI on top if you want. OCSNG has clients that run on your machines, and check back to the server every so often, we use 1hr intervals. This isn't a problem as we only have it do stuff every 24 hrs, and you could of course increase this if you wanted.
With OCSNG you get a great database of what's on your machines - what hardware, OS, service pack, software, and even username (if someone is logged in when the inventory runs). So right away you have all this info programatically gathered. It also supports deploying software to Windows and Linux (though we only use it on Windows right now for that).
The nice thing about it's deployment technology is it's a pull from the client. So no firewall issues or insecurity on the remote machines. They use SSL to authenticate the commands, though the download is in the clear. And it works great on laptops, because if the client isn't on the network when you set the deployment, no problem. It just picks it up the next time it's on and starts the install. Same for desktops if users have incorrectly turned them off at night.
This is great for a current snapshot of the machine, but what if you want to track changes? GLPI is the answer - it integrates with OCSNG and reads in the database every 5 mintues or so. GLPI also supports software license tracking, network layout modeling (though this is entirely manual)- help desk ticket tracking + knowledge base/FAQ and plugins for more.
GLPI isn't as mature as OCSNG (which is at v3.0.1 or so even if it's NG v1.0.2 or so), it's at 0.68.3 - but mostly it's entirely useable for hardware/software inventory, location etc tracking. I expect you could use it as a helpdesk ticket tracking as well, but we already use RT so...
Finally, you might say - this is great, I know what I've got out there, where it is (with some data entry) and I can push software. How do I track if it's actually on in near real time though? What do I do with my switches etc?
Well, the answer is ZenOSS Core.
Zenoss.com
Pretty easy setup, great community support, and full commercial support if you really want to spend money. This does full SNMP monitoring of v1 & v2 SNMP. But so what, lots do SNMP... It also supports WMI monitoring, so the same box can monitor our Windows Services! This is big. Plus, it does nice graphing for performance monitoring if you use the free SNMP Informant on Windows, and set up your permissions on your linux boxes. It scales to 4,000+ machines if you've got the RAM. It supports multiple distributed monitors(not that you'd need that unless you have a truly huge network).
It does full e-mail alerting + event reaction, that is you can have it fire off an e-mail, and have it restart the service remotely + e-mail you the problem has already been fixed! You can build really complex alerts and hirearchies of notifications if you want.
If you're looking to update or possibly change how your deployment and monitoring works, FLOSS is looking more and more attractive with the responsive communities, open standards and of course free and commercial pricing and support.
Now, I haven't had the ability to compare various commercial offerings beyond looking at their documentation as each is quite expensive. But for those of you without oodles of dollars, I have identified, and used with some success some FLOSS tools. Some of these are really quite mature, some are usable with great potential. For some reason, many of these seem to come from France.
I could talk for some time on various tools, but for now I'll just lay out what I was looking for, and what I've (well, the team I'm on) have done.
We wanted something that would work across Linux, Windows and MacOSX. More is great, but those were important - we wanted to consolidate our management tools, not increase them. We already have management for each platform separately except for inventory of some sort, so that's were we started - inventory. And there is a great package of two projects that does this nicely, and has a useful package deployment tool as well.
OCSNG + GLPI. The "names" for the two projects.
ocsinventory-ng.org
glpi-project.org
So, you have to have OCSNG - you can layer GLPI on top if you want. OCSNG has clients that run on your machines, and check back to the server every so often, we use 1hr intervals. This isn't a problem as we only have it do stuff every 24 hrs, and you could of course increase this if you wanted.
With OCSNG you get a great database of what's on your machines - what hardware, OS, service pack, software, and even username (if someone is logged in when the inventory runs). So right away you have all this info programatically gathered. It also supports deploying software to Windows and Linux (though we only use it on Windows right now for that).
The nice thing about it's deployment technology is it's a pull from the client. So no firewall issues or insecurity on the remote machines. They use SSL to authenticate the commands, though the download is in the clear. And it works great on laptops, because if the client isn't on the network when you set the deployment, no problem. It just picks it up the next time it's on and starts the install. Same for desktops if users have incorrectly turned them off at night.
This is great for a current snapshot of the machine, but what if you want to track changes? GLPI is the answer - it integrates with OCSNG and reads in the database every 5 mintues or so. GLPI also supports software license tracking, network layout modeling (though this is entirely manual)- help desk ticket tracking + knowledge base/FAQ and plugins for more.
GLPI isn't as mature as OCSNG (which is at v3.0.1 or so even if it's NG v1.0.2 or so), it's at 0.68.3 - but mostly it's entirely useable for hardware/software inventory, location etc tracking. I expect you could use it as a helpdesk ticket tracking as well, but we already use RT so...
Finally, you might say - this is great, I know what I've got out there, where it is (with some data entry) and I can push software. How do I track if it's actually on in near real time though? What do I do with my switches etc?
Well, the answer is ZenOSS Core.
Zenoss.com
Pretty easy setup, great community support, and full commercial support if you really want to spend money. This does full SNMP monitoring of v1 & v2 SNMP. But so what, lots do SNMP... It also supports WMI monitoring, so the same box can monitor our Windows Services! This is big. Plus, it does nice graphing for performance monitoring if you use the free SNMP Informant on Windows, and set up your permissions on your linux boxes. It scales to 4,000+ machines if you've got the RAM. It supports multiple distributed monitors(not that you'd need that unless you have a truly huge network).
It does full e-mail alerting + event reaction, that is you can have it fire off an e-mail, and have it restart the service remotely + e-mail you the problem has already been fixed! You can build really complex alerts and hirearchies of notifications if you want.
If you're looking to update or possibly change how your deployment and monitoring works, FLOSS is looking more and more attractive with the responsive communities, open standards and of course free and commercial pricing and support.
WIDGETIZE