Claris moving away from CentOS... going forward with Ubuntu

This sure got my attention this morning:

Deprecation and removal of support for CentOS Linux from Claris FileMaker Server

5 Likes

One of these days, someone will finally get it that continuity, reliability and low cost of ownership are the primary objectives for the majority of business information systems.

I am so glad we didn’t rebuild all of our cloud servers to move them from Windows to CentOS. I guess this will result in the third change to FileMaker Cloud?

I do feel for all the system administrators out there.

4 Likes

I think CentOS was selected specifically for that I imagine (continuity, reliability and low cost of ownership). That decision was taken some time ago now, in a context different from the one we know today. We get to look at this after the fact, but if Claris had chosen Ubuntu on day one, how many of us, with the information available at the time, would have claimed they would have been much better of with CentOS for its continuity, reliability and low cost of ownership.

Red Hat (acquired by IBM in 2019) simply changed how they do things, and they announced it in December 2020. Could Claris stay on CentOS instead of going to Ubuntu? Could they adopt something else than Ubuntu? Probably, but it seems their call in this new landscape is that the best approach is to go with Ubuntu.

Red Hat's decision has ramifications on a lot of enterprise production system and everyone has to adapt, not just Claris. Not everyone will come to the same conclusion, but that is up to every decision maker.

It all comes down to cost vs risk: The best safeguard against risk is diversification, but diversification has its own cost. Centralization and specialization tend to lower cost, but increase risk.

There are recent events that clearly warn us about the dangers of relying too heavily on one single element for any given purpose:

Today I see Claris as diversified, allowing the client to run on Windows & Mac (with separate products for iOS and web deployment, more recently partnering with Livecode to add to its reach) while also allowing the server to run on Windows, Mac and Linux. Time will tell if they stay diversified or get tempted by cost reduction strategies.

Today, just as it was yesterday, no one has to "rebuild all their cloud servers" and that is a good thing. The ones who do, do so by choice, not because they "have to", but because they seek something they, by their own decision-making process (flawed or not), cannot get with diversification. That is ok. Like everything, it has its pros and cons, implications, ramifications and consequences. It has its own risk. Risk is a cost that has not materialized.

3 Likes

I agree today. However, according to the documentation

FileMaker Server 19.2.1 will be the last version that supports CentOS

If FileMaker Server 19.2.2 is released tomorrow, then that won’t be the case unless people are happy to use older versions of FM Server.. Users of the original AWS FileMaker Cloud have to find the time and skills to move to the current FileMaker Cloud. It is reasonable to assume that the next FileMaker Cloud will not be built on CentOS there will be more changes, which are unlikely to be a transparent migration for the current clients. I hope I’m wrong.

I understand the risks and the unknowns, I’ve over 35-years of this, but for the majority of organisations their IT is not their raison d’être, it something that should just work for them while they focus on the really important issues relating to their business. Constant change is nothing more than a distraction and a drain on precious time and money, which is even more of an issue in the current state of the world.

1 Like

MBS Plugin will work on Ubuntu, too.

6 Likes

Original AWS FileMaker Cloud is a separate thing. The customers were given notice that the product was being replaced and they can choose what they want to move to (on prem, Claris FileMaker Cloud or other cloud offering from a 3rd party host). Claris replaced that product with another one, and simply for the amount of noise there was with customer being exposed to the AWS side of things, I am very glad they rolled out a newer implementation where AWS is not customer facing. If a customer is able to run their own AWS instance, then they can still do so on supported OSes.

This is not so much about diversification vs low cost, it is more about early adoption vs opting for a deployment that has been in market for a while.

The ones who went from AWS straight into their own implementation of CentOS are indeed suffering and feeling they get forced into back-to-back changes. I simply believe there are not that many of them out there.

Not everyone is running FMS on CentOS. A number of them are running it on Windows or Mac. CentOS was made available to everyone with FMS 19.1.2, released on October 28th. The news from Red Hat, issued on December 8th, called for prudence. So, I'm simply trying to keep things in context and say that we are really talking of a window of less than 45 calendar days. For people who would have planned a migration of a system that is otherwise working, or deploying a new system altogether. How many of them actual production systems? I don't know, but the numbers can't be that high. I guess my question for you then becomes what should have Claris done to avoid what you call here "constant changes".

As for Claris FileMaker Cloud itself, I have high hopes Claris will be able to migrate things in a way that is transparent for the clients. I'm not quite able to say how likely it is and would be curious to know why you think it is unlikely. You say "the next FileMaker Cloud", but I'm not even certain Claris will position itself that way. I think they will change the backend and customer facing elements will remain the same, requiring little communication about the change itself. If Dropbox changes from CentOS to Debian, if they go from AWS to Azure, I'm pretty sure that won't trigger much from their communication side of things.

Even things with long term support will come with changes. In 2021, constant change is pretty much the name of the game, accounting for security improvements and other improvements. "Set it and forget it" does not really exist anymore, peace of mind on that side of things can only be bought by asking someone else to perform some work for us and that usually takes the form of service contracts or subscription services.

But yes, what we all want is for it to "just work". It just so happens there is a lot of people who make a living in making things work and keeping them working.

Just for the record, because I assume Claris, running their own servers on Linux, will only have more data available to them on the usage that is specific to that platform vs Windows or Mac to inform decisions that can shape the product, it will likely make me favor linux deployments more and more, over time. I understand that everyone's decisions can follow a different reasoning and reach a different conclusion.

Robert, if they can do the migration transparently for their customers, then there isn’t a problem. Separating the data and configuration from the system should make this possible. If people have built their own CentOS on prem or VM, then this will cost them time, disruption and therefore money.

I’m not trying to make this into a lengthy post, I just feel it seems such a short time since FM Server for Linux was released and it is all change already. The rights and wrongs all depend on whether you look at it from Claris’s, the end user’s or the money earner’s point of view. We tend to always try to look at it from our customers’ point of view.

Kind regards
Andy

2 Likes

This is not a Claris issue. This was a Redhat issue. It affected, and will affect for a while, many deployments across all platforms and deployment models.

Consider what would happen if Claris was on an annual release cycle... MORE people would have had time to move over to Linux and CentOS, and then be forced to change. This is better.

And lastly... CentOS vs Ubuntu on a headless server... what difference does it make? Yes it's different, yes, there are some pieces that will require different lines of code to automate... but it's a one time change.

This isn't a pattern of changes that will continue for years on end... it's the result of using open source... anything. You are at the mercy of the developers and community, and you take that risk of the developer deciding they need to monetize it in order to maintain it.

3 Likes

Yup. Agree.

1 Like

Understandable. But when you build on something kind of shaky, don't be surprised by surprises. I am talking about Linux. I know that many IT people love Linux, I understand. But the large choice of distributions, with all their differences - just think about their different package managers - makes it hard for Linux to be on the side of continuity. Another example: you want to install MySQL on Linux, the installation depends on the distribution, up to the directories used. And if you upgrade your distribution, tough luck, the installation will need to be updated.

Last year CentOS world was shocked, will Ubuntu or Debian, or XYZ be the next one ?

Speaking of continuity, who would have predicted that Apple would abandon the Server app - I know it's not needed for FMS. But for all those companies that are Mac shops, that was bad news. Looks like Microsoft is alone in the continuity.

Speaking about the Exchange nightmare, seems many users were not up to date with the patches and that is not the way to go.

2 Likes

Different installers for different OSes is just the norm, and it results in different deployments (folders, prefs, etc.). Windows does not have that many flavors running concurrently (not speaking about support timeframe here) like with Linux.

Let's compare apples to apples please. If you take MS Windows for comparison, please compare it to a paid OS on the linux side of things. My bet is Red Hat does pretty well for the comparison you are attempting to make concerning continuity. Closed source vs Open source is a different debate altogether.

Apple has been moving away from the server space for quite some time.

Concerning Exchange, even if it may be true that a good number of servers are not up to date with patches, MS rushing to release "emergency security updates" does not sound like the problem was because people were not patching their servers. I do not think downplaying something like this is helping anyone.

From this other article:

Microsoft credited researchers at Reston, Va. based Volexity for reporting the attacks. Volexity President Steven Adair told KrebsOnSecurity it first spotted the attacks on Jan. 6, 2021.

Adair said while the exploits used by the group may have taken great skills to develop, they require little technical know-how to use and can give an attacker easy access to all of an organization’s email if their vulnerable Exchange Servers are directly exposed to the Internet.

“These flaws are very easy to exploit,” Adair said. “You don’t need any special knowledge with these exploits. You just show up and say ‘I would like to break in and read all their email.’ That’s all there is to it.”

I'm simply quoting from the article. I myself think the quote is a bit of an exaggeration, but one to reflect how, once aware of the flaws, exploiting them is easy. Given how it is from the very person who first reported the attack, I'm willing to let them choose the wording. There were 4 security holes that required NEW patches. People being affected by this has nothing to do with lagging behind on patches. Applying security patches is good practice, simply could not save you from this one.

I'm not trying to say "MS is bad, Linux is wonderful" nor the opposite. I think what I reported is factual, you can make your own choices from there. I was simply pointing to diversification vs non-diversification each having their own pros and cons.

Providing FMS based on a FOSS OS is all about continuity, reliability and low cost of ownership.

CentOS is a rock solid OS and a great choice as a platform when it was made. We've all been wondering which OS they would move onto next. There are a lot of good options. Ubuntu is another excellent choice.

The provision of FMS Linux is just good news all around.

1 Like

FYI, our hosting service fmcloud.fm is ready for Ubuntu. Actually Ubuntu is even better from our point of view.
So if you're using our service or planning to do so, don't worry, we will be able to make the switch on day 1 of FMS release. (well, that is if you need to, because we usually prefer to wait a few days before upgrading FMS, just in case…)

I would expect you offer a seamless automated migration to a new VM with Ubuntu, right?
And on success switch DNS to point to new one.

As seemless as one can be with FileMaker (as you know, we still have to close the databases to ensure data integrity).
But one of the many advantages of docker over VMs is that there won't even be a change of IP address, so you don't have to wait for a DNS propagation. Yes, it's so cool! I'm still not used to how cool docker is :slight_smile:

So just to make it clear, once the FileMaker process is stopped (which can take from 10 seconds to a few minutes -that's the same regardless of the hosting type), it takes about 5 seconds to relaunch with the new OS. All in all, for 90% of our customers, the 'migration' is done in lesss than a minute, with no IP change or DNS propagation.

6 Likes