Slashdot Mirror
VMware CEO: OpenStack Is Not For the Enterprise
coondoggie writes "VMware CEO Pat Gelsinger says he doesn't expect open source cloud project OpenStack to catch on significantly in the enterprise market, instead he says it's more of a platform for service providers to build public clouds. It's a notion that others in the market have expressed in the past, but also one that OpenStack backers have tried hard to shake."
Big corporate CEO says open source projects are only for geeks, children and people who can't afford it. News at 11.
I'm pretty sure that CEOs have been feed this so much by their marketing executives whose paychecks are on the line that they truly believe it. It just makes it so much more fun when they file bankruptcy or get bought out and the new company cans them without their golden parachutes.
"Be particularly skeptical when presented with evidence confirming what you already believe." -
I work for a cloud storage provider and have vSphere and OpenStack clusters and there two are for different tasks. The 'fighting' over the two is comparing apples to banana peels.
He may be right that it's only public cloud adoption *now*, but we (enterprises) are looking at the following as our 3 years road map:
The big problem we have right now is that it's hard, if not impossible, for us to take our big, giant, poorly design monolithic application into the public cloud. We need to implement the cloud methodologies and characteristics internally (elastic, scalable, on-demand etc) before we migrate that compute to a pay per cycle model.
In three years time when we've done the above - I can only imagine how much more stable and mature OpenStack will be.
He may be right, right now.
That's because people have historically coded their apps with the assumption that the database/hard drive/web server/IP address would always be there to write to or read from. They're also written with vertical scalability, i.e., if things are slow then throw faster hardware/more IOPS at it. All of these criteria vmware is good at handling.
People are now writing simple apps that use ridiculously complicated frameworks to ensure things work even when they're pear shaped. Most of those apps are written so scalability is horizontal. More speed comes from throwing more hardware at it. This also increases reliability.
These are usually done by new startups because they have specific needs (avoiding paying a SAN vendor) and skillsets (coders who don't understand, or don't know about the availability of a hardware solution so they code something in software.) The thing is, yesterdays startup is tomorrows enterprise. They won't migrate away from whatever cloud stack platform they're running without serious thoughts to the problems it may cause.
I'd guess one of the reasons a vmware CEO would say openstack isn't a competitor is they're owned by EMC, a SAN vendor.
Having said that, we evaluated openstack for our business and didn't like the rough edges in places. We're using a mix of vmware and proxmox right now.
Look at e-mail as an example. Globally and between corporations people have long used free/open standards, protocols and applications, sendmail, smtp, postfix, etc. However a growing number of users are moving from stand alone e-mail clients to web based e-mail platforms such as hotmail, yahoo mail, gmail, and so on, each of which have the option of being accessed through stand alone clients, or through their web interfaces.
When you enter the corporate environment you largely switch to comercial web server and clients. Perhaps most often Exchange and Outlook, respectively.
That said, many compaies are using open source platforms as their interface to the rest of the world. Whether that server is between firewalls in a DMZ, on some external service provider is irrelevant.
Similarly tremendous portions of internal corporate networks are running Microsoft web servers to host content internally, and managing content with Sharepoint. While there are some examples of each on the Internet, most corporate public interfaces and a the vast majority of other available servers are open source / free Apache, and other servers, with open source php, postgres, python, and so on backing it up.
Based on that model, VMWare and Zen instances will be widely used within corporate environments, however I strongly suspect that OpenStack will be largely used on the Internet in general.
The hazard with saying it will only be used by 'hobbiest' and 'geeks' is that when you get down to it, two of the largest entities on the Internet today, namely Google and Facebook, were started by hobbiest and geeks. And both started with free/open source, software, and are largely using that to this day. In other words people experimenting with new ways of making the Internet work for them are going to do so using the resources they can get the most value for their dollar from, and that's far more likely to be OpenStack than it is VMware or commercial instances of Zen.
You never know...
The usual FUD reaction when they see an open source technology compete with their core business. Free Hypervisors made them lose money on just providing those. Now they need to get the money from the enterprise management system tools they made. Unfortunately, open source tools try and manage them all, while their business is based on managing mostly their own hypervisor offerings and not the open source ones, or the ones from their competitors.
RedHat is in on OpenStack and they're putting big bucks behind it. Give it a few years, and VMWare will be the one that has to catch up on Enterprise readyness. Just managing a single group of KVM or XEN hypervisors is already working just fine if you use RHEV (the paid and supported version of oVirt) and I have no doubt that managing clouds will be on par shortly now that big money and many developers are being deployed.
I was promised a flying car. Where is my flying car?
We already have a large VMware installation - and due to the way our business works (customers work with us and the servers we provide for them for years instead of months or weeks, almost no "peak load" stuff that requires dynamic provisioning...), I feel we don't really need a scale-out platform (which OpenStack seems to be) but rather a virtualization-platform.
If we were to implement OpenStack, we'd have to build in parallel:
Add on top of that the fact that it usually requires a lot of time and effort to get anything built "right" (and seldom on the 1st attempt), I doubt we'd make a lot of savings over VMware even in the medium term.
Even more concerning in my view is the fact that most of the corporate "backers" of OpenStack sell public "Cloud-Services" themselves - we have already learned the hard way (via a different "cloud" product) that when for these companies the need to choose between customers of such a public service and those with a "private cloud" installation arises, they will most likely tend to favor their public cloud customers (or whichever business is bigger).
Coupled with that comes my prediction that OpenStack will "fragment" rather sooner rather than later, with each of its backers offering some sort of "enhanced" ("enterprise") version (with stability patches and some additional features) that may or may not be a bit cheaper overall than VMware (all things taken into account), leaving you with a solution that works "almost like VMware, for almost the same price".
Am I too pessimistic?
Windows 2000 - from the guys who brought us edlin
Xen (which the ISV be loving!)
That's his point. Xen (or the commercial XenServer for that matter) has had very very little enterprise penetration. Even when it is used in the enterprise, it's usually not the primary hypervisor but isolated to a specific application within the company. VMWare currently owns that space and the only short term threat to them is Microsoft and Hyper-V. Service providers, however, love it because their primary concerns usually revolve around costs (thus why you see commodity hardware in a lot of cloud provider data centers vs Cisco/HP/IBM/Dell in the enterprise space).
I browse on +1 so AC's need not respond, I won't see it.
No, the cloud, in the end, will eat most corporate IT. 20-30 years down the road most companies won't have internal IT anymore. Like everything else these days, corporate cost cutting will eventually win out. Sucks but I can't see it playing out any other way.
BR Most of the current instability can be attributed to industry growing pains. They will work the kinks out and the weaker companies (by ability,not necessarily size) will fall away.
I browse on +1 so AC's need not respond, I won't see it.
So regarding Openstack we already moved away the "first they ignore you" phase and going somewhere into "then they laugh at you, then they fight you", uh?
Interesting.
I agree for the small environment. A 50 man company will not need 1-2 IT people.
For the Enterprise, I doubt it. If you have more than 1,000 users you still have enough on-site hardware and networking to worry about, that you'll still need IT. Even if it's just for making purchasing decisions, pushing buttons for off site support, information governance, project management and resolving issues when multi-cloud services are in play, it's the latter that becomes the problem.
If you have multiple tenants for the same cloud services in one company behind one IP address, things start getting really interesting really fast. As when you and a partner company use different cloud services and you start having difficulties. Trust me.. Microsoft / Google etc. can't just stick Wireshark on their data centre to work out exactly what's happening. All current troubleshoot tends to rely on a non-cloud partner having these skills to give them the detail that they can't afford to investigate.
I'm not a nimby, I've helped and encouraged the use of a number of cloud services in my enterprise, particularly for easily solved problems or where it's so specialised or small that it's really not worth anyones time learning how to do it in house. I'm sure there will be a move to outsource a load of services to cloud providers.. It'll be when people try to switch for the first time that the difficulties arise. So I expect "Cloud Reconsidered" in 3-5 years. Probably also if there is another major 2E2 debacle or security breach or outage.
I cloud be wrong.
Jason.
Hell, I run a esx host at home, even if its just for 3 perm linux servers, and 2-3 dev linux servers for fun, lets you try/test new distros quickly.
Even if its just for the snapshot feature, its worth it, considering fedora is so flaky at updates, its cool to have that 'undo' option with one click.
Liberty freedom are no1, not dicks in suits.
you guarantee you'll lose. MySQL isn't for the enterprise. Linux isn't for the enterprise. Hadoop, SaaS offerings like Salesforce, GDocs, etc. Once you say something isn't for the enterprise you almost guarantee that it will be a > $1B business (or fundamental tech used in a $1B business) in the next couple years.
But I don't think Open Stack supports anything equivalent to High Availability, where a VM will automatically reboot on a different host if it's current host goes down. If they could do that, I'd highly recommend everyone sell their VMware stock.
These two features are the heart and soul of the value in VMware's ESX product. If and when Open Stack can do both of those things, you'll have 90% of what you really need in a VM environment.