Xenserver 7.3: Changes to the Free Edition
xenserver.org
xenserver.org
I get the non-profit,school dilemma. and ideally, they do something about that.
but they aren't closing the source of their system ( everything is remaining open source ). They are removing some value add that for the most part, only business users care about. ( to the binaries they provide, are they forced to give you free compile time and bandwidth too? )
For example, I myself have been guilty in previous companies of spending $1M in hardware, and have used the free version of xenserver ( 120 huge vm node deployment, hybrid cloud ), to great success, and Citrix as a company has gotten zero of that.
So here I am, an user extremely happy with software that is being used to run a multi million dollar, profitable business, spending $8-15k per server, and giving zero to the company that made part of this deployment successful.
I gave hundreds of thousands to dell ( and indirectly intel, ram manufacturers, etc ) I gave hundreds of thousands to Arista ( including support and software licenses )
but zero to Citrix, because they gave me such a great free product, i didn't need their support.
License fees have certainly been out of whack with many of these companies ( per socket licensing for example ), but the fact remains, they are providing a business value, and should be compensated. ( and no, ovirt isn't quite there yet, vmware is, but that's even more expensive ).
Not sure what the answer is here, but I do feel that if you are a money making business, buying servers for thousands, a percentage of that should go to the software you use to make those servers useful.
I think the modeling of being paid is fair and businesses should work towards paying something. Even then though, it's tough because licensing adds up. Some of the startups we mentor that are running XEN have very similar setups to the non profit. Many are running older EOL Dell servers with multisocket. One startup we work with has 5 servers in a rack, all purchased for under $1500. With the standard pricing agreement from Citrix, that would work out to $7,630 per year. This is all expected from a completely bootstrapped (pre-funding) startup.
I think it's a bad faith move removing features, I understand pushing the support need, but that can be addressed in a different manner.
Totally respect where you're coming from - but at some point isn't that kind of their problem? It's not your job to monetize it for them.
I respect them making this move, however I think they had better options. Part of their problem is it's TOUGH to go from giving something away to monetizing. At the same time the people like you who were making a profit off of it probably will begrudgingly pay for the upgrade.
I think they should've gone the Redhat route - where they only release source to patches between releases. If you REALLY want to apply the code and build the new binaries yourself, so be it. If you find value in them writing and providing the patches - pay for it.
There is no perfect answer, but I think the people calling for their heads are being a bit silly. I have no doubt the two options were: make this change, or kill the project entirely. ESPECIALLY in the face of Amazon moving away from Xen - I would imagine there was some back-end funding going on there.
-
Citrix used to have XCP ( Xen Cloud Platform ), as a way of OSS the core functionality away from the paid functionality. They moved to one build in the possibly naive thought that making licensing a feature of a support contract, would enforce a social contract of "hey, why not pay these guys".
That did not work.
I think AWS moving away from XEN is a bit moot. While AWS did contribute back some to XEN, they mostly forked it for their use case. They are doing the same with KVM. Citrix has been by enlarge, the majority contributor to XEN.
Putting the sources on GitHub is something, but only with a very complicated build process which is not documented or even using URL's that are private/internal to Citrix.
That's why people start to push Citrix to at least release a documentation on the build process and/or exploring the difficulty to build it. See https://xen-orchestra.com/forum/topic/562/xenserver-7-3-and-... and also the Citrix thread regarding that: https://discussions.citrix.com/topic/391868-xenserver-73-ann...
So far, no answer from Citrix.
There is a way to do it, you just don't want to figure it out. You want to FORCE them to figure it out for you. BooHoo.
Again, this is a commercial product, that they open sourced. Their commitment was to release the source. They then additionally went thru the step of providing free services ( free binaries, open support system, etc ). This has not worked out for them, they also need to run as a business. They are still giving you ALL of the source.
RedHat went thru a similar period, and in all honesty, was the only thing that saved them as a company.
If a Centos projects comes out of this, awesome. if XCP ( Xen Cloud Platform ) comes back as a volunteer project, awesome too!
"Opensource" is about many things. But to me, and many others, one of the early core tenements has been that the users of a product help out in some way. It doesn't mean you get to whine your way into others doing free work for you.
GPL compliance requires you to provide build scripts.
There is a gap between not posting a doc and having no way to build it because it uses internal Citrix URL to the dependencies. I'll will try to figure it out by myself, but I suppose you have no idea regarding the huge task it is.
This is not a normal situation: you can't brag you are open source but avoid people to build it themselves (that's what happening actually). You need to be clear about that. I have no problem if they remove the sources, at least the message is clear. But telling it's Open Source without providing something you can actually build IS the problem.
According to legend (I wasn't actually there - this is secondhand, so do take with a grain of salt), they tried to convince a company I worked with previously that they needed an additional license for the server that processed backups. No Oracle client or server installed; the logs were pushed via ssh. When they balked, Oracle tried to increase the price two more times. It was apparently great motivation to move to MySQL.
Maybe that was just one particularly bad rep, but I don't think I've gone through a single year of my career without hearing something absurd about them.
There are more choice quotes in someone else's HN comment about Bryan Cantrill on Oracle.[2] I highly recommend any talk you can find of Bryan Cantrill. A wonderful mix of industry history and great delivery.
It's not pure capitalism, which would provide mechanisms for dealing with this, and it's not too much regulation, which could also prevent this, it's just the right amount of both that allows Oracle to play them against each other to game the system.
Also, their (re)licensing process for existing customers pretty much amounts to extortion. You better be really sure about how much it would take to rewrite your systems without Oracle, have a plan how it would be done, and advertise it to them - because as soon as they'll feel a lock-in, prepare to open the wallet for ridiculous sums for trifling things that you've been doing for years for free, but now their lawyers can find an weird interpretation where that requires extra licenses; they don't even attempt to increase the revenue by providing some extra value or service for you but simply by applying their muscle.
I sat outside the door of the CIO's office. He wasn't expecting anything shocking and so left the door open during the call. Oracle had previously told us that we wouldn't have to pay for licenses for non-prod instances. When they went back on that promise, the CIO said (well, shouted) "Okay, fine! Tell me how much you're going to fuck me. As soon as you're done with that, I'll fuck you right back by throwing Oracle out of my datacenter!"
A contract was signed and during early implementation Oracle EOL'ed a significant module with a replacement that was planned but did not exist. Followed by attempts to charge us for their development of the replacement which, again, had been a pre-paid module in their off-the-shelf ERP. Already paid for.
That didn't fly, but later the replacement lacked key features: A web-facing portion collected information. We demo, it works, we ask where data processing folks view it in the forms app... crickets. Conversation ensues:
O: "It's in the database"
Me: Sure, but where do users view it when collating the other information?
O: "Well, a dba can get to it"
Me: Okay, but when we go live, will it show up as a another tab in a form or...?
O: "We've completed our build of this feature"
Me: Hold on, collecting & processing this data is a critical and contractually itemized deliverable. You have developed half of that deliverable.
O: "We've completed out build of this feature"
I was probably fortunate that my CIO was there, and took over at that point. A few weeks later, Oracle walked off the job. There had been delays in a modules I wasn't involved with, and Oracle wanted more money. It was a fixed-price contract, and we wouldn't re-open negotiation on a signed contract already multiple milestones in on its payment schedule. Oracle walked. Literally gone, no word, no message, just dozens of people that were there on Tuesday, gone on Wednesday. That's Oracle.
Xen has been dying for a while. EC2 is heading off it, and they've held out the longest. This change is evidence that Citrix would like to accelerate that trend by killing off the home lab users that stuck it out (read: me). Looks like I have a new Christmas project.
https://cloudplatform.googleblog.com/2017/01/7-ways-we-harde...
https://www.theregister.co.uk/2017/11/07/aws_writes_new_kvm_...
https://blog.linode.com/2015/06/16/linode-turns-12-heres-som...
https://www.digitalocean.com/help/technical/general/ (Cmd+F KVM)
But if you’ve got multiple hosts oVirt is great.
1: https://www.joyent.com/blog/why-smartos-kvm-dtrace-zones-and...
Nice simple web GUI.
1: https://medium.com/@MARatsimbazafy/journey-to-deep-learning-...
I wrote up my approach at [1], in case you're curious.
The purpose of the base image is to preinstall something that's ready for NixOps to "take over" with whatever I actually want to run on the VM.
VMWare has very little OSS, especially for their equivalents of Xen server. Id support KVM and co before VMWare just on those grounds alone
EDIT: clarity
Couldn't care less about open source. I'll prefer something that's free and works perfectly without hesitation.
- Dynamic Memory Control
- Xen Storage Motion
- Active Directory Integration
- Role Based Access Control
- High Availability
- GPU Pass-Through
- Site Recovery Manager (Disaster Recovery)
- XenCenter Rolling Pool Upgrade Wizard
- Maximum Pool Size Restricted To 3 Hosts (existing larger pools will continue to work, but no new host joins will be permitted)
That's quite a list.
I mean, seriously the entitlement is astounding to me. OF COURSE they're going to make it difficult to build on your own, that's part of the monetization strategy. If you don't want to pay for it, and you also don't want to spend the time figuring out how to build it on your own, don't use it. But to act as if they owe you something is... ridiculous.
Trying to make that sharing onerous is a violation of those principals and its not entitlement to point this out.
Its a astoundingly stupid monetization strategy. I'm going to assume that building the software may in fact be challenging but its a minuscule fraction of the difficulty of constructing the complete solution.
So imagine it takes 1 person who is paid 120k a year a few weeks to produce a viable solution and documentation. The cost of this solution is about 2k and scales to infinity people.
In a universe where nobody communicates or collaborates I suppose you could imagine that everyone would pay 10k instead of paying 2k in labor for an officially supported solution perhaps in consideration for other value provided like tech support.
The problem is that this fails to consider the fact that someone might actually donate this labor for free to everyone negating the benefit of this particular moat.
It also creates a situation where someone might be inspired to put in the 2k worth of labor and charge 100 users $199 each and come out ahead.
Its certain that a number of players have earned money selling some sort of value add on top of open source software but the value add really needs to be something that you have a competitive advantage at providing not something trivial like a difficult build process.
You are defending nonsense.
Really? Because that's exactly what Redhat does and they've been continually held up as the shining example of "open source works!" for several decades now...
I experimented with the 7.2 Free edition and right from the installation not succeeding at first to many other things not working as expected it was not a very good experience. The management client UI is also Windows only I think and it's very sub par.
Given the new changes I wonder anyone would have any reason to run xenserver free edition. There's always ESXi, HyperV and good ole KVM (which maybe an issue if you run Windows server as a VM.)
EDIT: The correct answer was posted by blinkingled before I finished typing this up hahah, read their post instead of this one :)
I'm afraid I don't have a holistic answer, in this instance, that satisfies the requirement that the developers eat and be clothed, and that the users have an unimpeded work flow. Maybe it's a moral solution needed between all users, of all payment tiers, to agree not to abuse the intent of the permissive nature of the free-tier user agreement. It will be interesting seeing how this plays out.
"I think it's important to differentiate between the pricing of the product from Citrix, and the source code licensing model. XenServer (all editions) remains open source, with the ability for anyone to head over to Github and peruse the code, contribute patches, and so forth. What's changing is what features Citrix puts into the free edition that it builds, tests, and maintains; all of those are still open source, though."
The binaries as provided by citrix come in 3 editions. These changes apply to those editions. These are 'supported' by citrix.
I guess there might be a reason compiling to distribute binaries would be a time-consuming pain? (Or compiling yourself, for that matter).
Makes me wonder if that gives Citrix an interest in opposing deterministic binaries.
A few months ago they started charging for patches and someone asked in the XenServer forums here:
https://discussions.citrix.com/topic/390349-open-source-fork...
The solution here is for citrix to roll back time, launch a new supported product, and then slowly close the faucet of new features going into the free product. They would have then created an opportunity to pitch the new innovations and migrate their existing customer base to the new platform. The only problem with this thinking (besides the time machine), are the competitors that are releasing more innovative products at the same "free" price point.
Companies, please learn from this: Anything given for free has no value until you try to take it away.