On 28.01.2021 14:54, Neal Gompa wrote:
On Thu, Jan 28, 2021 at 6:22 AM Gena Makhomed gmm@csdoc.com wrote:
What is about running in the one bare metal RHEL server virtual machines with different subscription owners? For example, run in production on one bare metal server 16 VMs with subscription owner Alice, and 16 VMs with subscription owner Bob, and 16 VMs with subscription owner Carl, and so on. Are such configurations legal and allowed or not? I didn't find any limitations on the blog article, but for sure and for future I need a clean and unambiguous answer from Red Hat.
If such configurations are allowed - this is a legal workaround for a limit of 16 no-cont RHEL instances. For example, a small company, with 50 employees can absolutely legally have free and no-cost 800 RHEL servers in self-support mode. Company with 100 employees can have 1600 free no-cost RHEL servers in self-support mode and so on.
If such configurations are forbidden (on what basis?) - I have no choice but to migrate from free CentOS and no-cost RHEL to Oracle Linux or Alma Linux or Rocky Linux.
And in the future if my company grows and I will need to buy commercial support - I will be forced by Red Hat's decision to buy subscriptions for Oracle Linux from the Oracle Corporation?
Is this the real goal of the no-cost RHEL 16 instance limit - force CentOS users migrate to Oracle Linux?
Brian Exelbierd explained this whole thing quite well on the Ask Noah Show[1]. The answer to this is that what you're saying is perfectly allowed. The bet here is that this is sufficiently costly, risky, and a hassle (who wants to manage 100+ Red Hat accounts? I know I wouldn't!) that the company in question would decide to purchase RHEL subscriptions from Red Hat, especially after experiencing the value that Red Hat provides (Red Hat Insights, live kernel patching, etc.).
But did you know the minimal price of one RHEL Server subscription?
~ 350 USD/year.
So, subscription for 100 servers/VMs will be cost 35_000 USD/year.
Every year. For 10 years price of 100 subscriptions is 350_000 USD.
You don't need 100 Red Hat accounts, for 100 server subscriptions.
For 112 RHEL Server subscriptions you need only 7 Red Hat accounts.
16 * 7 == 112.
Managing 7 Red Hat accounts really is sufficiently costly, risky, and a hassle? I don't see any problems with such work for 7 accounts.
For 1600 servers minimal commercial price is 560_000 USD/year. and price is 5_600_000 USD for 10 year subscription. For 1600 servers.
Managing 100 Red Hat accounts really is sufficiently costly, risky, and a hassle? This work cost more then 5_600_000 USD for 10 years?
I don’t understand one thing, if it is so easy to get workaround these restrictions of 16 no-cost RHEL instances and at the same time bypassing these restrictions is completely legal - why were these restrictions introduced at all?
So that CentOS users should think about whether they should switch to no-cost RHEL, or maybe they should think about switching to completely free variant of Enterprise Linux from Oracle, which does not have such restrictions on the number of no-cost instances and don't need any subscriptions for seamless work at all?
As previously CentOS Linux users live (mostly) without commercial subscription and support from RHEL, the same in the future, they can live with no-cost Oracle Linux (mostly) without commercial subscription.
If user of mass installation of Oracle Linux in future need commercial support - he/she will buy commercial support from Oracle Corporation, not from Red Hat/IBM. It's obvious, isn't it? Money will go to Oracle.
This is some kind of strange situation when the Enterprise Linux was created by Red Hat staff and Fedora community, but the Oracle Corporation will make additional money on it, because this is where a large number of CentOS users can/will go in current situation.
What I would have liked to see is the addition of some generic low-cost subscription options that would be sufficiently below the floor to fit with even low-margin businesses so that as a business grows from 16, to 50, to 100, to 1000, and so on, the company would continue to use RHEL and continue to support the awesome work Red Hat does. Right now, the current pricing is so unbelievably expensive that I would instead just convert the boxes from RHEL to CentOS Stream after a certain threshold.
CentOS Stream is just a beta-version for next minor RHEL release.
CentOS Stream is not ready for production, see for example, bugreport https://bugzilla.redhat.com/show_bug.cgi?id=1913806 - this bug is present in CentOS Stream 8, but absent in CentOS 8.3.
I firmly believe that low-cost self-support options would be a good value for Red Hat to offer to the market, especially for a lot of those startups that eventually grow past the 16 server limit. I hope that's on the docket based on the comment at the top of the RHEL blog post that this is the first of many new programs.
I hope so too, because if they do nothing, then many CentOS users will simply leave for Oracle Linux. I just don't see any other way out now.