OpenStack private clouds are SCIENCE PROJECTS says Gartner OpenStack can run a fine private cloud, if you have lots of people to throw at the project and are willing to do lots of coding, according to Alan Waite, a research director at Gartner. Waite works at Gartner for technical professionals, the arm of the analyst aimed at hands-on IT pros rather than suits. His opinion is …
I'm always suspicious of Gartner when it comes to cloud computing, I have never get the impression they really understand what its all about.
TBH their comments about OpenStack could be applied to ANY private cloud software (proprietary or open source). There are very few successful platforms mostly because the guys designing them (usually with the "advice" of Gartner et al) never think through the actual use cases, so end up building a platform that has no value to end users because they are locked into the usual IT department echo chamber. It becomes yet another "transformation" massive project that's all about jobs for consultants and ultimately changes nothing for the actual users.
Public cloud platforms (by which I mean AWS) are successful because they targeted the developers and application code and made it easy for them to deploy and run applications. With that in mind there are plenty of platforms that give the users what they want (maybe not what IT think they want, which is the problem) and don't require a huge amount of integration or effort.
OpenStack is actually one of them and there are enough distributions out there now that are easy to set up and use (at least for core functionality) that I wonder what research Waite actually did!
I'm no analyst fanboy, but of all the analyst firms, Gartner actually has the best grasp of cloud and what is real IMHO. For one, they actually use the different products in addition to talking to clients about their experiences. Try asking an IDC analyst about the last time they actually used a product they "analyze" and see them splutter.
OpenStack has some unique problems that are *not* shared by other cloud software due to the structure of the project and the intensely competitive vendors and many competing agendas. Proprietary software has the benefit of vision, focus and a coherent agenda that is utterly lacking in OpenStack, and software companies collecting money from customers actually care about and spend engineering time on things like the install and upgrade experience. With the open source model, this is the stuff that no-one wants to work on, so they don't, and you get the crappy OpenStack deployment and upgrade experience.
Your comment about how easy it is to set up shows that you haven't done it beyond a trivial case (just one server, perhaps?) Sure, there are good single node OpenStack installers. But how about a 100 node deployment? With virtual networking? And identity integration? With Swift? And some useful services? That's hard. Just ask anyone who has tried it or done it. You actually know less than the Gartner analyst you're trashing.
It's hard to judge what he's talking about without more context, but the bit about "it is not a cloud management tool" seems to be the main point. Yes, it isn't a management tool and it isn't supposed to be. The management tools are provided by other people. For example, if you use Ubuntu's OpenStack implementation, then you use their management tools. It's too early in "private cloud" evolution for the management tools to have become a commodity yet.
As for "working with the stack is not straightforward" is sort of missing the point. Building your own Linux OS by downloading the sources from the various developer repos isn't straightforward either. Perhaps that's why most users don't do that. They use a ready made distro from Red Hat, Suse, or someone else instead. The same will be true for OpenStack. Pick a vendor supported release, don't build your own unless you are Amazon or Google.
And "Gartner sees mainly large organisations trying OpenStack"? Wow, big organizations use software meant for big data centres. Who would have thought that? Next Gartner will be telling us that IBM mainframe salesmen are having difficulty penetrating the lemonade stand market.
Meanwhile, el Reg has another story detailing how Microsoft is thrashing about with different versions of Azure trying to find something that will work in the private cloud market. They're running into the same problem that everyone else is in this market, which is trying to define some actual benefits for smaller customers.
For the average user, any cloud system, OpenStack included, is something which someone else puts together for you. It's sort of like the next level of operating system. Eventually this sort of thing will percolate down to smaller scale systems in push-button form, but possibly with something else (e.g. Docker) layered on top of it.
At the moment however, it would probably be a good idea for people working in the IT industry to download some of the free "cloud" systems and give them a spin to get familiar with the technology first hand.
For once i think this gartner report seems good. I'm sure there are a ton of management types out there that think if they install openstack they have an instant cloud.
I keep getting updates from a friend that uses it and the word remains steer clear. IaaS is not important to me we get along fine without it. Too much hype behind openstack. Like SDN too much hype there too.
Happy with vmware and "utility computing". Cloud can go to hell.
“OpenStack is great as an open source standard for infrastructure access,” Waite told the Gartner IT Infrastructure, Operations & Data Center Summit” in Sydney today. “It has great APIs. But it is not a cloud management tool..."
All we need is for some enterprising open source folks to come up with a suite of cloud management tools on top of OpenStack, and this Gartner analysis gets turned on its head.
In 2011 I set about deploying a strategic BI platform - everything from server tin to fancy front-end tools. I was challenged to use Openstack at the time, but I didn't feel it was ready. Ended up with Intel, VMware, Sharepoint etc, loads of capacity for new projects, sadly no chargeback capability, but looking back, I still think it was the right decision.
Regardless of the spin that Gartner analysts provide, there are some key tidbits of information which certainly can be corroborated. Take this little nugget: "Waite said the firm has counted just 740 implementations anywhere, “because the use cases are pretty small.”"
This is exactly what we found when we started working with OpenStack. In fact, we could only find one use case...to provide "less expensive" infrastructure for Platform as a Service, in our case Cloud Foundry. So if you look one step deeper, the real use case is twelve-factor application delivery. And that is still something that is very new. Just look at the total number of PaaS deployments.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
