Convenience trumps 'open' in clouds and data centers Call it OpenStack. Call it Open Compute. Call it OpenAnything-you-want, but the reality is that the dominant cloud today is Amazon Web Services, with Microsoft Azure an increasingly potent runner-up. Both decidedly closed. Not that cloud-hungry companies care. While OpenStack parades a banner of “no lock in!” and Open Compute …
I can sympathise with David Laube's position.
Sadly, my experience of the "open" projects I've dealt with, specifically open-source, is that documentation is lacking.
It's an old saw, but there is a certain amount of truth in the saying "open source is only free if your time has no value"
On the "install it and use it" side, the Nagios documentation claims to be updated daily (http://library.nagios.com/library/products/nagioscore/manuals/), but apparently hasn't been updated since 20 September 2013 (http://nagios.sourceforge.net/docs/nagioscore/4/en/).
On the development side, I recently had to debug some OpenSSL issues, and the API documentation is a nightmare. I have yet to find any decent API documentation for the Linux kernel. I've looked at https://kernel.org/doc/htmldocs/kernel-api/, and compared to e.g. MSDN, both the OpenSSL and Linux kernel API docs are a very pale imitation.
I'm not knocking the end results of any of the above - good bits of software all of them (recent OpenSSL issues aside), but I suspect the problem is that documentation isn't fun, and the people best placed to write proper low-level technical documentation, those with the deepest knowledge, are the least likely to want to do it.
I'm with you on this.
I was looking at the pie chart and it struck me that companies fail with the private cloud because of "Failure to change" in proportion of about 44% yet when it comes about the public cloud, they do change without any problem and all it is a big success with big hugs and pink unicorns dancing all across the rainbow. Anyone can share some wisdom with us on this matter ?
If you could point me in the direction of those unicorns I'd be grateful as the company recently opened a unicorn meat packing factory but have found unicorns that we aren't able to capture any unicorns for some reason, Management insists they're right around the corner though. Bullshit on the other hand is everywhere... If only we'd decided on a fertilizer business.
The problem with any open environment like this is that, even with controlled release schedules, things are going to change quickly and there is no guarantee anyone is going to keep the documentation up to date. Note that there is no guarantee in the closed source world either, but VMWare, Microsoft and others do offer paid support and fixes for problems that you don't have to roll yourself or ask very nicely for someone else to.
If you're Facebook, Amazon, Google or any other company driven primarily by technology, than you will most likely be able to demonstrate the case for keeping a team of OpenStack experts, Linux kernel hackers or any other open source stuff you use. If you're an academic institution, you have staff whose job is ostensibly to learn and use new technology, and contribute to its development. If you're a company who only uses technology, you will most likely be happier with VMWare or Microsoft handling things for you.
Businesses have willingly gone in for lock-in forever. IBM's cash cow is still companies paying by the MIPS for mainframe access, and in some cases, there's no where for that workload to go without causing a major disruption. Microsoft has had the Windows/Office lock-in on most businesses for a very long time. What this model does offer is stability and accountability, which businesses like. Open source projects really have evolved lately, but most businesses are still not ready for a "roll your own" system. Decision makers are especially leery about support being provided on a best effort basis. Look how much Red Hat makes by effectively wrapping open source Linux in a support model.
Interesting article but I think you missed the real driver behind the public IaaS adoption, developers!
Anyone who is using AWS in a big way is most usually a developer who went there because their own internal IT platforms were slow and a bit rubbish. Developers couldn't give a crap about the platform they run on, all they want are programmatic API's, instant response, and the ability to get on with their job. So your absolutely right, its all about convenience for the developer.
Couple that to the fact that most private clouds aren't even clouds, VMware on a bit of tin (even if its "converged") is not a cloud, show me a programmatic API, then I'm listening. The complete shambles of traditional vendors and SI's pushing "private" cloud that's been happening over the past 3 years has helped drive people out.
Now for the "but", there are numerous organisations who whilst starting in public have come across the big problem. Cost! When your application hits mainstream and you have a nice base-load your suddenly hit with the realisation you are paying an awful lot, especially when finance start crunching some numbers.
That's when people start to look at building internal clouds, difference is they now have to build platforms that look and feel to their developers like a public platform (i.e, API driven) and that's when the lack of open becomes a big problem. Ever tried to move an AWS app on-premise?
So now the conversation about open platforms becomes important. Its also true that customers are smart about not being locked in, first "open" Unix killed the mainframe, then x86 killed Unix. Developers themselves are mostly using open source tools and technologies so it makes sense to have the platform open as well!
By the way the whole "OpenStack is hard" argument is a bit old and quite frankly smacks of a lack of research, there are plenty of decent supportable, install-able distributions out there now that means a customer doesn't get their hands dirty and have to worry about how it works.
It's a bit too early to make judgments on private clouds yet. The technology is still in the development stage. Computing technology tends to follow a path of
1) experimental and unreliable.
2) proprietary, unique, and expensive.
3) proprietary, common, and less expensive.
4) open source, commodity, but uses proprietary add-ons to make it easy (open core).
5) open source, commodity, free, and painless.
We're still somewhere in between stages 2 and 3. We won't see benefits from widespread private clouds until we get to stages 4 or 5. Until then, we need the proprietary management systems and the experienced staff of companies like Amazon to handle it for us, or else you need to hire your own expert staff. The larger IT using companies like banks are doing the latter.
Projects like Docker and others are working on taking us to stage 4. Stage 5 will come some time after that.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
