Wednesday, November 2, 2011

New developments in CentOS world...

Suddenly, there is a flurry of news about CentOS. First, there was some lengthy discussion on CentOS user's mailing list about the current state of the 6.1 release and about the RedHat versus CentOS. Also, on a development mailing list there is discussion about moving CR repository back to main repository. Finally, there was a question on Slashdot about why would someone use RedHat when there is CentOS. So, let me give you a digest of some of the topics that were discussed.

RedHat Company vs. CentOS Project

Relationship between CentOS and RedHat often comes under discussion and suspicion. And with recent changes that RedHat made, some of which made harder for CentOS to follow RedHat, discussions become even more energetic.

It is true that it bacome hard for CentOS project to follow RedHat releases. The reason is that RedHat on purpose made this process more complicated than necessary. That's why people are bashing RedHat assuming that it is complicating things because of a CentOS. But, there is high probability that RedHat is doing this because of Oracle and Novel. Oracle is building it's own distribution based on RHEL and then it is selling services for it. Both, Novel and Oracle, are selling services for RHEL too, and with lower prices than RedHat itself. And Novel and Oracle have much more resources on their disposal than CentOS will ever have. So it seems logical that RadHat is not concerned with CentOS, but with Oracle and Novel. If it were concerned with CentOS than it would be a major tactical error.

Not only that, but CentOS developers are repeatedly stressing that CentOS without RHEL, and RedHat, wouldn't exist, and if you need, or can, that you should buy support from RedHat. This is not done by either Oracle nor Novell. It turnes out that buying RHEL service isn't so expensive, after all, as I'm going to discuss shortly a bit later.

Also, there is information that someone from the core team said  that RedHat helped them, but these claims are unverifiable, i.e. I can not find mails where this is written, or anything that could be used as a proof. But, one member of the core team told on a mailing list that they were contacted by RedHat to verify that they (the CentOS team) are aware of Acceptable Use Policy (AUP). It is interesting to follow the tone of this particular discussion. It started as an information that RedHat contacted them and they asked CentOS if they are aware of new AUP. Quickly, this tread turns into bashing RedHat. Finally, the developer in question responded himself. And the conclusion was that it was not hostile neither there was any threatening! The developer again explicitly states that CentOS tries to obey AUP as much as possible.

Finally, from some informal discussions that some users had it turns out that RedHat is looking at CentOS beneficially. But I somehow doubt that RedHat can or will help CentOS, at least not publicly. And may I say that whoever expects from RedHat to do so, isn't expecting the right thing to happen. Instead, those should expect CentOS project to change itself.

As a side note, I didn't know that Microsoft officially supports CentOS, but only as a virtual guest. This is quite interesting.

CentOS Project

It seems to me that the main problem to CentOS is CentOS itself. First, for such a large project more people should be included in the core development team. Second, the lack of communication from core team is also very seriously hurting CentOS. At the time this post was written there was an announcement from Sempember, 1st that only 16 packages have to be built and 6.1 will be released. But it was two months ago, and in the mean time there wasn't any announcement or status update. People that use CentOS are those that need some stability and predictability, and both of those seems to be missed by CentOS. All this makes people look for alternatives, and more frequently than not Debian and Ubuntu popup. I hope that I don't have to explain how this hurts RedHat in longer term.

And while I'm at alternatives, it was rather interesting to find out that there is also additional alternative to CentOS and Scientific Linux based on RHEL, the distribution neatly named PUIAS.

RHEL Pricing model
Part of the discussion were concerned about licensing models. Someone wrongly calculated that you have to pay for a RHEL license $2000/year for 2 sockets (two places for CPU). But then it turned out that there are cheaper options, namely, buy only OS without support services. In that case you are paying $50 for a workstation per year, and 350$ for a server per year. These options are actually very attractive for people that use CentOS for a simple reason that in both cases you don't have support, BUT you do have latest security updates.

Also, there is an option of paying $2000/year for virtualization platform with unlimited number of guests. Furthermore, this options covers RHEL licences in all guests. Obviously, in case you plan to virtualize some other OS, then you'll have to buy licenses for that OS separately.

CR Repository

Apparently, there are also discussion to change how transitions between point releases is done. CR (or Continuous Repository) has been introduced with the idea that security updates are faster released. Even though some are happy with CR, while others are not, it seems that it isn't good enough (TM) solution. First, there are many older installations that don't have installed CR repository. They do have installed updates repository, and they require manual install. This is definitely problematic. Second, constantly some question or something like that pops out in some mailing list. The new idea is that point releases are fixed (e.g. 6.1) while generic release (e.g. 6) is a moving target. All in all, it seems that the change is inevitable but we'll see what exactly is going to happen.

No comments:

About Me

scientist, consultant, security specialist, networking guy, system administrator, philosopher ;)

Blog Archive