Slashdot Mirror
OpenStack: the Open Source Cloud That Vendors Love and Users Are Ignoring
Brandon Butler writes: "OpenStack has no shortage of corporate backers. Rackspace, Red Hat, IBM, Dell, HP, Cisco and many others have hopped on board. But many wonder, after four years, shouldn't there be more end users by this point? 'OpenStack backers say this progression is completely normal. Repeating an analogy many have made, Paul Cormier, president of products and technology for Red Hat, says OpenStack’s development is just like the process of building up Linux. This time the transition to a cloud-based architecture is an even bigger technological transformation than replacing proprietary operating systems with Linux. "It’s where Linux was in the beginning," he says about OpenStack's current status. "Linux was around for a while before it really got adopted in the enterprise. OpenStack is going through the same process right now."'"
The only people with the business case to use cloud infrastructure are the corporate backers themselves. SMB have no reason to chase clouds and mid-level B2B computing crap gets outsourced anyway.
Cloud computing companies fuck their customers with excessive charges orders of magnitude higher than normal data-center co-location costs.
The reason why people are ignoring it is because they recognize the ass fucking. Simple really.
If I want to host my own, I get VMware in my own datacenter. If I want to host in the cloud, I buy storage+compute from AWS. I see no reason to deploy OpenStack at a small to medium sized business. Am I just looking to get myself fired for insisting on a solution that is not VMware?
Thank you cloud to butt.
In the beginning, Linux was free. I remember using it in college and learning about it and getting excited. If these big corporate players want traction against AWS and the like, they should be giving out free hosting to college students so they can tinker with it too.
...that right now, in the midst of the NSA security nightmare and all the angst and FUD it's causing, that people are wondering why individuals are not deciding to throw their often-sensitive data into the cloud.
how could anyone think their data will be or stay safe, given the various threats that we hear about on almost a daily basis?
timing is everything (besides location of course...and sex appeal...and everything else) in life, and right now is not the time for cloud computing.
never bring a twinkie to a food fight.
...until upstream bandwidth in the USA catches up with the rest of the world, self-hosted "clouds" like this are just not happening. Sure, you can colocate a server, but that's expensive for a SMB and you can spend that same money on a bigger Internet pipe instead, but with such cheap turn-key on-demand scaling services like EC2, why set up your own?
The problem with OpenStack is that the only part of it that is actually open and free is DevStack, which is fine for tinkering but isn't suitable for actual deployment. Of course, for an actual deployment solution, you have to pay these vendors $$$$ at which point you might as well just pay Amazon $$$$ and get better service.
There's many more OpenStack users and operators than you think. OpenStack is good for small cloud vendors, people that want to run a private, in-house cloud. It's good for Universities that want to teach Cloud computing, or enthusiests that want to try setting up their own private cloud for toying with.
OpenStack holds a summit every 6 months. This last one (just last week) had over 3500 people in attendence - developers from those sponsoring it, operators, and user; and they were talking about how phenominal the growth has been - the first from what I heard had like 500 people.
So while you may want to use AWS, Azure, or Google Cloud Compute for a non-managed, public cloud; if you want to do something in-house, you have fewer choices. VMware certainly has their offering; but it also comes at a high price (yes, I've looked at it in the past). I'm not sure where the various hypervisor support is, but I do know they use KVM and have the ability to use others (Rackspace uses Xen, others use VMware or Windows HyperV if I am not mistaken; at the very least there's discussion on it).
Now, I wouldn't expect high growth for OpenStack. Why? It's a big budget item to run in-house, and most are probably not going to market they use it. If people are not devoting a lot of money up-front to run it, they may be testing and slowly rolling it out as resources allow. And yes, you can run it from the SMB level to the Enterprise level.
Disclaimer: I work for Rackspace; I've got a few servers that I may try to install OpenStack on to play with myself as well.
Truth is like the sun. You can shut it out for a time, but it ain't goin' away. - Elvis Presley (source: imdb.com)
Linux got adopted in the enterprise because it already had an existence elsewhere, bourne
out of frustration of failing proprietary software, lock in, closed cultures, fuck-over-ism, etc.
(MS are you listening?) The techies injected their own culture, knowledge and the need
for Linux into the corporations organically (no doubt this amounted to a revolution).
Cloud computing on the other hand is the answer to a non-exsitent problem. It's a coporate
hype. Adopters are those that've got their heads not in the cloud, but up there where the
sun don't shine.
Something about associating the word "Open" with a cloud service makes me uneasy.
Hell, lets even ignore all that. What would I, as an end user, use OpenStack for? I'm sincerely asking: what is the use case for an end user directly using OpenStack?
It's not gaining traction because there's another huge movement going on at the same time, developing into a useful deployment product far faster then OpenStack.
It's also much more approachable for individuals and small teams.
Docker/LXC is moving fast and is quite amazing. It's no-frills VM's with consistent/predictable templates with absolutely minimal overhead.
Why run OpenStack to virtualize hardware into sections, and then run many heavyweight OS VM's on slices of hardware, when all you really wanted to do was isolate your web app or database process with its own CPU, Memory and Filesystem.
Docker is a far better technical solution for application deployment covering concerns and most use cases for networking, cpu, memory and disk isolaton.
I'm enjoying openstack for the ability to develop / demo free node/mongo stuff for freeeee. That said I'd most likely bite the bullet and go AWS if cash flowed.
OpenStack is more about virtualisation than anything else. Its potential usefulness for "cloud" service providers is one example, but it's probably of more interest to large organisations looking to consolidate their own in-house IT services. As with many "open" technologies, the realities aren't quite as simple as the article here suggests, though.
It's certainly true that proprietary high-end networking gear and virtualisation software can be expensive. In that respect, alternatives like OpenStack are potentially disruptive.
On the other hand, ask anyone who's actually had to administer an OpenStack system how they feel about it, and the response might be a string of curse words that would make your mother blush. This is a technology (or more accurately, a loosely connected family of technologies) still very much in its infancy, and sometimes it shows.
Also, just because big name brands are keen to be associated with the shiny new buzzword, don't mistake that for sincere support. OpenStack poses a direct threat to the established business model of some of those networking giants, and just like everyone else, the executives at those businesses are wondering where the industry is going next and how to look like you're playing nicely while really still trying to optimise your own financial position.
If you disagree, post your argument. (-1, Overrated) isn't your personal censorship tool for views you don't like.
I thought UNIX/LINUX were a hardcore part of enterprise groups pretty much from the start.
I'm not sure what they mean by "end users". I've been keeping an eye on OpenStack, and it seems to be useful for developing cloud applications, but I might be missing the point a little. So are we calling developers of large-scale applications "end users"?
I don't think people are ignoring it, but as far as I know, OpenStack doesn't really service your standard network IT market yet, and it's not really something that will service "end users" as I think of them. It seems to be something to provide scalability for development, but if you're a developer working on a large application, it's often smarter to go with a vendor rather than trying to build your own infrastructure. That means that they go with AWS or Rackspace or something.
So my question is, who do you expect to be implementing OpenStack other than cloud providers (e.g. Rackspace) and a relatively small number of companies looking to build their own cloud infrastructure?
As an IT guy (not a developer), the whole thing is still pretty unclear. What would I use OpenStack for? If I wanted to test it out, what would I need to get started? How would I set it up? What, then could I do with it? Most of the appeal of "the cloud" at this point is the potential to divest myself of responsibility for its maintenance. The only people that I can imagine making use of OpenStack are large companies with large public, business critical web applications, and even then only those who, for whatever reason, don't want to use AWS or Rackspace or some other vendor, and have the resources to build and maintain a bunch of cloud infrastructure. Yes, there are businesses that fit that description, but it's not a large percentage of businesses.
And I'm not sure I'd call them "end users".
To really make use of the cloud, don't put traditional apps on it. It is not designed to run things like MS Exchange.
If you work in a software development shop, especially a web app, then cloud is awesome. Think of cloud as an API. That is where the real power is!
We have a continuous integration and continuous delivery pipeline. The entire deployment is described in software using Amazon's API. We abstract our infrastructure as code so we can replace it with Openstack if we need to. Amazon's API far ahead of anything else out there, so right now we don't really need to switch. This system is extremely powerful. We can bring up entire testing environments the the execution of a script. In system configuration is driven with Chef, but even some of those scripts use the Amazon API to help discover information about the environment.
VMWare provides some of the features, but nothing like Amazon offers. VMWare is also designed for a traditional IT cycle where you can about running a VM for more than a year. Cloud thinking makes more using of disposable nodes. A machine may not last a month because it is replaced with an entirely new image.
So, IF you write software correctly, having an in-house cloud API is extremely useful. Having a cloud API that a standard is also very useful. Start small with a public provider (Rackspace), then bring in-house as the business grows (RedHat Openstack). When the business needs somethings more elastic, that same API can be used with third party providers to supply the computing when it is demanded (Rackspace).
Cloud API's are new. Give it time.
Hell, lets even ignore all that. What would I, as an end user, use OpenStack for? I'm sincerely asking: what is the use case for an end user directly using OpenStack?
That, I think is the answer to TFA's question. You shouldn't. That's for the DevOps people to worry about.
As an end user, you shouldn't have to care what the data center underpinnings are. And for personal systems, the standardized images of cloud systems aren't much use. Although if you're running a call center or some other group where lots of people are running essentially identical systems, they're a better candidate for commodity virtual hosting.
Ive been looking to overhaul our IT department past weeks and upgrading all our "physical" and loose KVM machines into a nice solution. I have looked around on all the possible options, besides openstack and apache cloudstack there is not many real open source options that are ready for enterprise use.
The openstack setup is very complicated if you want to make a redundant, load-balanced setup that will give you real 99.99% uptime as required for any enterprise. I'm up for the challenge but I guess the documentation could be much better and adaptability will grow
Those that mention NSA first read what openstack is before calling it "cloud" (as in public cloud).
I used them for years,finally gave it up.
The API was not good enough, and for my use, cheaper VPS providers were cheaper.
"Piter, too, is dead."
Having played with OpenStack for a couple of weeks, the overriding impression I got from it is that it is made by vendors, for vendors.
You can download and do most stuff advertised, but it takes a LOT of learning and work, UNLESS you pay a vendor for their bolt-on to make the management of it easier. What a surprise, most of these vendors are the developers. It's therefore in their best interests to make it difficult to setup.
What tipped me away from it was that once you have it setup, it is also going to require a significant amount of maintenance to keep it going (adding new VMs, maintaining users, etc). For a medium size setup (perhaps 16 physical hosts), you are looking at a full time job for one person. Once you start seeing that, VMWare looks a lot more attractive and has a lot more features on top of that, not to mention is tried and tested, rather than what feels like a beta platform.
I'm not usually in favour of commercial, but OpenStack needs a lot more polish before it is useful for real production work, without the cost of vendor lock-in.
Is what the summary aludes to: 95% of the people I see who are 'in' Openstack are not users, but people assigned by vendor 'X' to make sure that vendor 'X' is not rendered irrelevant. A large chunk of the resource behind openstack verges on technical marketing rather than development.
I see this as more worrisome than the Linux case. Linux adoption was also developer heavy with few users, but developers with genuine passion were on it. Here we have an ecosystem of vendors that is fearful of 'the next linux' and putting armies of developers on it to push agendas around as much if not more than push actual technical capabilities. There are some passionate 'true believers', but by volume you mostly have developers doing it as 'just another job'. Linux has certainly coped with that, but only after a very long period of baking in an architecture before the vendors got motivated. Openstack got slammed with vendors on day 0 and thus the whole architecture is afflicted with some pretty gnarly stuff and I'm not seeing a lot of signs that those will be addressed.
Thus far when I see openstack implementation start in earnest by a site, it evolves within a year to either being given up or being Openstack in name only as they just replace most of it with home-grown tooling that works.
It's a big budget item to run in-house
And this is one of the issues with it. It doesn't quite manage to make things significantly easier than rolling your own stuff. It bears actually a resemblance to many vendor driven industry standards in this way: uselessly open ended so everyone's agenda could be accommodated.
XML is like violence. If it doesn't solve the problem, use more.
| Disclaimer: I work for Rackspace;
That's quite a disclaimer! By all accounts, Rackspace will be acquired soon (like well before the end of the year). It's going to be purchased for its customer base --not it's NASAnine technology. Check financial press (search CNBC, etc.) for specifics.
As far as AWS, Azure and Google --they all have lock-in APIs and, with the exception of Google, their future is not certain. "Cloud costs are coming down" --is BS. What these semi-literate "enthusiasts" posing as bloggers should be writing is that "Cloud prices are coming down". If anything, the "cost" of the cloud is going up --it's a function of land and utilities, yo. As far as going into your boardroom and saying "we're going to be moving our servers to Google", good luck with thatæ
VMware has vCloud-providers (and it's own vCloud service too), but who knows who the Hell those people are running those services. In at least one case, I found a jackass who bought a datacenter after the dot-com bust and has sort of adlib'ed his way. As much as vCloud is a convenience for VMware users, it enables jackasses like that guy to dig a very deep hole for himself and his customers.
"The Cloud" has been hyped out about as far as you can go. For most corporate enterprise environments the stock configuration of a "shake and bake" standardized instance does not cut the mustard. That breaks the whole cloud model when it comes to enterprise workloads. For people who try to force it to work they end up getting lost in a sea of hundreds if not thousands of configurations that were used only a few times. Factor in the effort to create those configurations and the effort to navigate through them and it's just easier to not use the system at all. These systems are great when you have a standardized environment. Not many companies have ever been able to accomplish that even with years of trying.
The other thing that really smart people fail to notice is the hardware. Once your "private cloud" starts to grow you start to reach the limits of the capabilities of the hardware. With nobody playing traffic cop to identify important applications once you run out of gas your locked into adding more hardware that you really did not need to buy in the first place.
Cloud technology works well for "disposable" companies that will get sold off in a few years or other not so critical work loads but for big corporations it's not a useful model. Some of these big companies are still using "big iron" because it works for their purposes. Companies don't really care about what the latest craze is. They have business goals and look at the profit and loss. For many companies the cloud fad will come and go without them even getting involved..
The vendors of course love the cloud craze because it's a concept driven by hand wavy developers as a way of saving costs and of course everyone wants to save costs right?. You have to admit it's a brilliant idea if it actually worked in the real world. Once you start looking at the model with a critical corporate enterprise viewpoint there are too many huge shortcomings for it to actually work well.
The only way to make "The Cloud" work is to get buy in from everyone and standardization. With all of the crazy certification and security stuff going on there's a huge financial reason why many vendors don't want standardization. They want a reason to sell you a support contract and often throw oddball technical stuff into the software to make it so you really need that support. These vendors will not let "The Cloud" eat their lunch.
I work for a research group doing work on agricultural simulators. I'm more of an applied mathematician rather than an IT guy, but I have a strong IT background. So when we received a bunch of old BladeCenters IT was tossing out we took them for use in our small cluster. We have Condor on them and they're fine, but I was thinking we could do more with them and I thought OpenStack would be the way to go. After reading the insanely long setup instructions (where the pre-install docs are almost longer than the actual install notes) I attempted to follow them to install... something... I don't know; I'm not a Linux admin so maybe it's just me, but there were so many modules and dependencies that I wasn't really sure what I needed. In the end I gave up, it was taking up far too much of my time for what amounted was really an interesting but not critical experiment.
I know part of this, maybe even a large part, is due to my inexperience, but I found the install process extremely complex. Is it supposed to be like this? Are Linux admins the only ones who are supposed to install OpenStack? Maybe if the bar was lowered a bit they might find more people willing to use it.
This whole cloud thing will blow over. As someone already noted, if you replace "cloud" with "other people's servers" it sounds a lot less appealing and a lot less manageable.
Do you want to outsource your outsourced infrastructure to a bunch of head-wobblers?
I want to delete my account but Slashdot doesn't allow it.
Well I have, and even with RedHat's documentation and distribution, it's nothing short of a nightmare.
It took me a good part of a day to subscribe to RedHat's evaluation distribution, and configure maybe 2 out of the 7 or so daemons that are needed to get it to all hang together.... and this was starting from scratch with no idea how the open stack architecture hangs together. In fact, I'm still a bit fuzzy on the details.
Compare that with a vmware ESXi install. Within an hour or so, you're running linux in a VM.
For a contractor going into an organization trying to sell this, it's very very hard. Skilled people in Open stack are few. I can't easily set something up in Open Stack and then walk away, or the customer is in a lurch for support. The technology needs to be well supported and well understood with a community of techs.
At the moment, while I love open source and everything you can do with it, a typical organization would rather go with vmware due to it's ease of use and the number of techs that can manipulate it. Yes it costs a fortune, but it's worth paying because it's easier to support, and these enterprises have money for this.
Openstack is going to go great guns where in-house techs can deploy it for customers, and spend all the time in the world to learn it's ins and outs....but for everyone else it's too much hassle.
The comparisons with earlier version of Linux are apt. Just as enterprises don't want to roll their own Linux kernel, much less do enterprises want to hand configure their own cloud.
There will be a market for preconfigured & value-added open stack environments however. It's just too early to call yet.
READY.
PRINT ""+-0
I do not use OpenStack because of their password requirements. I forget the details, but I remember trying something like 3-4 passwords, and each failing because of a different constraint.
Nopers. You can't give end users that hard of a time making accounts and expect us to stay (regardless of whether the password benefits are safer or not).
The data insecurity and NSA pre-packaging they offer is a big advantage but not for you
FTFY
I've fallen off your lawn, and I can't get up.
My experience as an end-user in a research project:
I've tried to install OpenStack on a small group of 4 machines (a controller, a network manager and two compute node). It was a real mess to install. The documentation contains omissions and mistakes. You need to write your own shell scripts to get the work done (and redone). Understanding what went wrong from the cryptic python debug messages is like banging your head against a wall. The only way I finally was able to test things was to scale back to a "one-node" system (everything on the same machine) and use DevStack. That works great but it's really far from a "cloud". You need to be HP or RackSpace to get this working well I guess.
Contrast that with OpenNebula. This platform is much less hyped about but it works much better. Even when you hit a bump on the road, you can actually understand the logs, and even debug stuff yourself. I got a 4 node system working with all storage on iSCSI and I can add more compute nodes seamlessly.
The thing about OpenStack is that it has been under really heavy development for the past two years. Two years ago the product was buggy as hell. But they've made a series of 6-monthly releases since then. Each one of which offered substantial improvements. Its now pretty good and stable. There is really a incredible support for it. I heard of numbers of around 2000 developers so each release really is substantially better than the previous.
Now that it is basically stable, it will likely get real traction with users and there are big private deployments already. The Australian NeCTAR project will roll-out 30,000 cores by the end of 2014. CERN is looking at a huge deployment of over 100,000 CPUs.
http://arstechnica.com/informa...