If you want to share devices between boxes, you'll want to check into GFS for linux (it's in the standard kernel) which will coordinate concurrent access to physical disks at the filesystem layer. GFS will work just as well on either a local (SCSI Y cable) disk array, or on a SAN, but the SAN disk will likely be easier to manage over time, especially considering you won't have to do anything strange to put more than 2 boxes in your HA configuration.
Also, linux-ha has STONITH capabilities (Shoot The Other Node In The Head) to prevent a split cluster from corrupting your filesystem(s).
As far as reliability, HP-UX (I don't think Linux can currently do this) can have dual paths to the same physical device on a SAN. This means you can pull the plug (or trip on the cable) on one of the fibre cards and the other will pick right up. It is possible that the qlogic cards will do this, but I haven't checked.
OpenView and Tivoli are for either very large budgets, or very large, distributed companies. NNM is pretty neat and all. It discovers your network for you, draws a really inaccurate map that you have to manually tweak. Then you find out that most of the features you want aren't part of NNM. You have to buy ITO (now called VantagePoint, IIRC). Then, you want to graph loads and network utilization. Guess what? Another $5-15k down the tubes. As far as I've been told Tivoli is the same way.
My point is that no matter which of those two you buy, you're going to need to do some substantial work to get them set up properly. Why not invest your time into something that is cheaper and, in most cases where you're monitoring
Where I work, I ousted OpenView and replaced it with NetSaint and Cricket. I also wrote a bunch of other CGI scripts to search my syslog archives and things of that nature. They aren't very difficult to maintain once you get the hang of it, and they're free.
If you're really set on something grand, I've been keeping an eye on OpenNMS which is more to the scale of NNM or Tivoli. Give their page a readover - they're nearing a 1.0 release, last I checked. Remember, you can always spend that cool million that's burning the hole in your pocket to hire the lead developer of one of those projects to come in to your company and 'Make it So.'
I just got a TS1000 and a TS2000 the other day. Great box - easy to set up and all that.
I would guess that the thing to do is either get a PPC that can compile compatible binaries (possibly running Hard Hat), or set up a dev environment for cross compiling. Montavista likely sells a development environment for their OS, so that might be an easy way to get started. Personally, I would try to find a powerpc that is similar and copy the OS off the cyclades for development. That way, you could minimize the surprises.
I started out programming MUD's on Windows 95 using Cygwin (which really sucked back then). When I complained on the list about compiler problems, somebody told me to try Linux. Slackware 3.4 made me into a Unix Admin. I installed it, screwed it up, fixed it, broke it, fixed it, et infitesimo. After a long time doing this, I said to myself, "Self, why not be a Unix Admin? You like it, you're pretty good at it, do it." So, I dropped out of college (3.5yrs Music School and.5 of CS) and started applying for jobs. No certifications, no degree - just my knowledge of Linux. My education was (and is) based entirely on the generosity of all the other admins out on the net.
In short, your best bet is to learn everything you can about unix. Even more importantly, learn just as much about networking and programming (C, perl, and shell) as those are just as important as your system knowledge.
Good Luck!
Slashdot Mirror
If you want to share devices between boxes, you'll want to check into GFS for linux (it's in the standard kernel) which will coordinate concurrent access to physical disks at the filesystem layer. GFS will work just as well on either a local (SCSI Y cable) disk array, or on a SAN, but the SAN disk will likely be easier to manage over time, especially considering you won't have to do anything strange to put more than 2 boxes in your HA configuration.
Also, linux-ha has STONITH capabilities (Shoot The Other Node In The Head) to prevent a split cluster from corrupting your filesystem(s).
As far as reliability, HP-UX (I don't think Linux can currently do this) can have dual paths to the same physical device on a SAN. This means you can pull the plug (or trip on the cable) on one of the fibre cards and the other will pick right up. It is possible that the qlogic cards will do this, but I haven't checked.
oops ... that hanging sentence is: where you're monitoring (less than)500 nodes, better.
that darn lessthan sign and html ... ;)
OpenView and Tivoli are for either very large budgets, or very large, distributed companies. NNM is pretty neat and all. It discovers your network for you, draws a really inaccurate map that you have to manually tweak. Then you find out that most of the features you want aren't part of NNM. You have to buy ITO (now called VantagePoint, IIRC). Then, you want to graph loads and network utilization. Guess what? Another $5-15k down the tubes. As far as I've been told Tivoli is the same way.
My point is that no matter which of those two you buy, you're going to need to do some substantial work to get them set up properly. Why not invest your time into something that is cheaper and, in most cases where you're monitoring
Where I work, I ousted OpenView and replaced it with NetSaint and Cricket. I also wrote a bunch of other CGI scripts to search my syslog archives and things of that nature. They aren't very difficult to maintain once you get the hang of it, and they're free.
If you're really set on something grand, I've been keeping an eye on OpenNMS which is more to the scale of NNM or Tivoli. Give their page a readover - they're nearing a 1.0 release, last I checked. Remember, you can always spend that cool million that's burning the hole in your pocket to hire the lead developer of one of those projects to come in to your company and 'Make it So.'
Good Luck!
I just got a TS1000 and a TS2000 the other day. Great box - easy to set up and all that. I would guess that the thing to do is either get a PPC that can compile compatible binaries (possibly running Hard Hat), or set up a dev environment for cross compiling. Montavista likely sells a development environment for their OS, so that might be an easy way to get started. Personally, I would try to find a powerpc that is similar and copy the OS off the cyclades for development. That way, you could minimize the surprises.
I started out programming MUD's on Windows 95 using Cygwin (which really sucked back then). When I complained on the list about compiler problems, somebody told me to try Linux. Slackware 3.4 made me into a Unix Admin. I installed it, screwed it up, fixed it, broke it, fixed it, et infitesimo. After a long time doing this, I said to myself, "Self, why not be a Unix Admin? You like it, you're pretty good at it, do it." So, I dropped out of college (3.5yrs Music School and .5 of CS) and started applying for jobs. No certifications, no degree - just my knowledge of Linux. My education was (and is) based entirely on the generosity of all the other admins out on the net.
In short, your best bet is to learn everything you can about unix. Even more importantly, learn just as much about networking and programming (C, perl, and shell) as those are just as important as your system knowledge. Good Luck!