Grid control is the way we manage large Oracle infrastructures, comprising many different databases on many servers -and, quite possibly, in many different server rooms. There is nothing stopping you from managing each database by logging onto it in SQL*Plus and tapping away your next SQL command… but it swiftly becomes impractical when there are dozens of databases to do it to!
The use of Grid Control provides a single point of management and control: visit it in your desktop’s web browser and you can control any number of servers and the databases that are running on them (it can manage middle tier servers, too, but that’s outside of scope here). Very convenient, very practical… fundamentally, it’s the only way to do it these days.
At the time of writing, the new 12c version is out (and the ‘c’ means we should refer to ‘cloud control’ rather than last season’s ‘grid control’!), but there is plenty of life in the 11g Grid Control yet: you’ll be installing it and using it for a good couple of years to come, I think, because businesses don’t suddenly switch to the latest and greatest of anything just because it has an exciting new version number.
In this article, therefore, I’m going to document how you install Oracle 11g R1 Grid Control (64-bit) and how you can use it to start constructing a larger Oracle infrastructure.
Before we start, know that this article expects you to know about, and to have created, a Kickstart server that is capable of performing both RCSL 5.x and 6.x O/S installations.
It will further assume you are not in a position to download Oracle patches and upgrades and that you want to experiment with this stuff on a completely free-of-charge basis.
Finally, I’ve written this article by running everything it describes on a desktop PC using virtualization with just 8GB physical RAM, a quad core CPU and about 60GB of real hard disk space. You do not need anything more substantial than that, but of course things will run faster and smoother if you happen to have a couple of spare PCs sitting around so that you can run everything on real, physical hardware. Either way, when I write about ‘building a server’, that can equally well mean creating a new virtual machine or pulling out an old Pentium 4 machine from the back of the cupboard, as you prefer.
Required Software
I have to say up-front that Grid Control is a bit of a monster to install, involving at least four separate software installations -and the patience of a saint. The complete list of what you need is:
- An operating system -an RCSL 5.x distro, 64-bit, of course. (Incidentally, it has to be a version 5.x distro, because the Grid Control agent never successfully completes its configuration on a 6.x distro).
- A Java Development Kit (JDK) -at the time of writing, I used version 6, update 26. You are on your own if you want to use anything newer than that.
- The WebLogic application server -the only version that works for this is the 10.3.2 generic.jar download. (It’s also known as ‘WebLogic Server 11gR1′ and you’ll need to click the ‘see all files’ link to expose the generic download in the ‘Additional Platforms’ column). Don’t try any of the newer WebLogic server downloads available on that page: they won’t work.
- The 64-bit Linux Oracle 11g Release 2 database software (Grid Control uses a database as its repository; that can -and ought- to be an 11gR2 database, even though Grid Control itself is called ‘Grid Control 11g Release 1′).
- The Enterprise Manager Grid Control software itself. You want the 11.1.0.1.0 version for 64-bit Linux. There are three files in the set, each about 1.5GB in size.
The Basic Plan
In principle, installing Grid Control is easy. You just follow the masterplan, which has these key features:
- Install a clean, minimal operating system and then configure it correctly for running Oracle
- Install Oracle 11gR2 and get a database running on it
- Install Java
- Install WebLogic
- Install Grid Control
All of this can take place on a PC which has around 30GB of free disk space and 2GB RAM, but if you’ve got more than those quantities available, Grid Control can certainly make use of them.
Getting Started – Installing the OS
Happily, this is one of the easiest parts to accomplish, thanks to the Kickstart server technology you will have implemented before reading this! You simply need to boot your Grid server with a network boot ISO of your chosen RCSL distro, then direct it to your kickstart script to finish most of the job without any user intervention at all.
Make sure you use the Diznix Kickstart Configurator to get a suitable kickstart script to begin with, and don’t forget that Kickstart can only work if there is a DHCP server available somewhere on your network.
Once the main operating system process completes, run the palmerston.sh shell script that you’ll find in the root directory (i.e., issue the command sh /palmerston.sh) to finish things off. Palmerston will ask for a username/password for the account that is to own the Oracle software (usually referred to as “the oracle user”, but you can supply any name you like), and will ask you for a default database name (I suggest something like “griddb” for the purposes of this article).
Here’s a set of slides showing you the entire process, from start to finish:
[nggallery id="4"]
The whole lot should take you less than about 10 minutes to complete -and once it has, you’re ready to get into the business of installing the Oracle 11gR2 software. So read on!
Very well done article Howard.
I’ve yet to use grid control, having learned to hate the existing DBConsole ever since it came out with 10G, I’m thinking it would only increase my level of frustration with it.. (I still install the java OEM on my 10G database servers, haven’t been able to find it for 11G yet) However, like you suggest, any DBA these days should get familiar with it, so I am going to take the plunge.
I have a couple of thing to think about though, before I take this on:
I can quite likely run GC from my workstation easily enough, I just got a new one and it’s fairly hefty workstation with a xeon 2.4 dual quad core processor and 12gb ram and a couple of TB’s on two drives, running windows 7 64bit. This however, would restrict me to the windows version, (which I’m assuming works just fine) but my preference is to have it on a separate machine.
I could either (a) create a VM within the virtual box app I have on my machine for GC or (b) use one of the physical workstations I have at my disposal, usually a P4 with 2-4GB ram, but it would be 32 bit. In fact I have one such machine with SuSE enterprise server 9 Linux on it running an Oracle 10Gr1 database on it I used as an experimental Oracle on linux test machine some time ago (we are a windows shop, until I can convince them otherwise)
I like linux, and don’t get the chance to use it as often as I would like, so installing GC on linux appeals to me. However at this time I can’t see this being much more than a one time install, so I’m more likely to just download an install distro iso file and install it locally. While your kickstart server looks appealing, I don’t think I really need it right now. Your recommendation for a 5.x install of one of the centos or SL distros is what I would do, but do these have any specific advantages or disadvantages over using Oracle Linux?
Were I to run GC in a virtual machine within my workstation, getting back to Oracle offerings, I see they have VM templates for grid control. I want the simplest path to having GC up and running as soon as possible, my last go around trying to install Oracle VM on a bare metal workstation in my spare time tells me your method is probably the preferred one here.
“do these have any specific advantages or disadvantages over using Oracle Linux?”
Big disadvantage: no support. (Whether that’s an issue or not, I’ll let you decide!)
Grid control doesn’t like encountering non-supported O/S: you can’t administer a non-supported O/S from it, for example. But that’s a matter of what O/S you run your database servers on and is therefore a separate issue. The actual problem is simply that if you run Grid Control itself on anything other than a ‘proper’ distro, you won’t get Oracle support should anything go belly-up.
It doesn’t have to be Oracle Linux, though: SLES is supported; so is Red Hat itself. Just not Centos or Scientific (which I always think is a shame, but is really quite understable).
I am jealous of your workstation specs, incidentally! (Apart from the Windows 7 bit, obviously!!) Still making do with 8GB and an ancient quad core here…
Can you please repair broken link to Palmerstone shell script
Very good tutorial congratulations
I may be mistaken, but I don’t think there’s any broken links in this article. There were, I think, deliberately no links to Palmerston in it, because it’s not intended that you would download it directly. You’re supposed to do a kickstart O/S build, using a kickstart script generated by my configurator tool. That script will include a download of the Palmerston script, but it’s *local*: you’re supposed to have built a kickstart server and put Palmerston on *that*. The whole point of Palmerston, after all, is to avoid Internet downloads at the point of performing an O/S install.
Of course, there’s then the issue of how you get hold of Palmerston for that kickstart server -but that’s covered in the article on how to build a kickstart server in the first place, and I’ve checked that that works.
If you *really* want to download the file directly, there’s a link under the ‘Downloads’ option on the site banner. I checked that works, too.
But, unless I’m much mistaken, I don’t think there’s a link to Palmerston in this article itself that’s broken… and that’s by design. (Give me the specifics if you spot otherwise, of course!)
Thanks for the kind words, too.