Slashdot Mirror
VMware: Hey, Other Cloud Services Exist
Nerval's Lobster writes "VMware has updated its cloud-management portfolio to support alternative tools, including Amazon's platform. That's a big step for the company, which for some time seemed to shy away from the idea of backing heterogeneous cloud environments. VMware's vFabric Application Director 5.0 is designed to, in the company's words, 'provision applications on any cloud.' That includes Amazon's EC2. The platform includes pre-approved operating system and middleware components for modeling and deploying those aforementioned applications, with the ability to use the platform's blueprints for deploying applications across 'multiple virtual and hybrid cloud infrastructures.' The other platform, vCloud Automation Center 5.1, enables 'policy-based provisioning across VMware-based private and public clouds, physical infrastructure, multiple hypervisors and Amazon Web Services.'"
It's quite possible that this move is in response to Microsoft building similar functionality into Hyper-V 2012.
VM There ??
I have yet to see a "cloud platform" that was much more than a collection of buzzwords.
Well, basically those "clouds" are just a cluster of servers running software which provides an interface where you can upload and run multiple VMs.
So for example if you want to run a website, you create a VM with the web server and data, upload it to the cloud service, and then have it run one or more instances to service HTTP requests depending on the traffic at the moment.
That makes it clear.
As vmware adopts ever more aggressive and unpalatable pricing schemes I'm not surprised that people are looking to cloud services.
Vmware makes some nice software, but they've been heading towards the "Well, how much do you got?" or "Give me your checkbook and I'll let you know how much it costs" or simply "Oracle-Style" pricing.
Glad they have some competition. The company that commercialized virtual machines is now facing the commodization of virualzation software itself.
It's just a fad. Buying hardware, putting a Linux distro in the CD drive, installing, configuring, driving hardware down to hosting center, installing hardware in a cage, testing, driving home, and then repeated all this when something goes wrong or needs upgrading is much more efficient. Nothing will even compete with a Linux server in a corner / basement / under a desk.
The details are still a bit cloudy for me.
"cloud is not about clouds"
"cloud means x"
"no cloud means x+y"
"yeah that's something we had before"
"it's marketing on old ideas"
"no, it's doing what we used to do, better, with newer tech"
"oh whatever my old tech was just as good"
"no it wasn't."
"yes it was"
"well now it's all cheaper and better"
This horse is thoroughly dead.
Amazon's offerings are pretty good. And yeah, their persistent use of acronyms is a bit annoying and confusing. It's not so much buzzwords though, as an attempt at branding.
Firstly, if you're just mucking around, it can be cheap. Really cheap. If you're just working on a proof-of-concept, you can possible get a dedicated server for free, as Amazon's prices scale on use, and a test machine isn't going to get a lot of traffic.
Secondly, it can be really fast. Because Amazon's physical hardware is already geographically distributed, you can do the same pretty quickly. If I have a properly setup application, and I decide I want a server physically located in Asia to reduce the latency to customers there, I can have it done in 10 minutes.
Thirdly, its easy. It's got a steep learning curve, but once you're on top of it, backups, disaster recovery, scalability and a whole host of other problems are essentially solved for you.
Sure, there's nothing it can't do that a properly configured and tuned geographically-distributed redundant cluster of linux boxes couldn't, but it's a hell of a lot cheaper, faster and easier than running such a cluster.
Just because you're paranoid doesn't mean there isn't an invisible demon about to eat your face
That's the Infrastructure-as-a-service offers. Then there's platform-as-a-service (e.g. App Engine, Heroku) and Software-as-a-service (e.g. Google Docs).
Essentially, cloud means: we abstract and automate everything from this level down, so you don't have to worry about it and can focus on everything above.
And yes, "cloud" is a buzzword for something that already existed. That doesn't mean the concept is bad or useless.
Dilbert RSS feed
Or is it merely "web application" virtualization?
I was surprised to read the Wikipedia blurb on EC2 to find they supported booting Windows 2008 images, I had assumed it was an "app engine" for some kind of web serving, and not the kind of virtualization you normally associate with VMware.
Does this mean that a company could theoretically run AD/Exchange from EC2? Skip over the usual hosting option and run it straight from there? I'm sure the pricing wouldn't be as good as a "pure" hosted Exchange solution per se, but you'd have the horsepower of Amazon.
Are people actually using it this way or is it generally only for hosting web servers?
I use the software service cloud for my personal and for work we have our DNS servers, web servers, load balancers, SQL servers, FTP servers all in the cloud. Works great, and has been an enormous money-saving option as our servers were at end of life.
This is very naive. You only talk of virtual hosting as a cloud. That's a 2006 state-of-the-art.
In fact, the GROWTH in cloud architecture comes as a transition of the traditional, Enterprise data-center. This is the next stage after the large-scale consolidation of Enterprise x86 computing onto a virtual platform for efficiency and cost.
A cloud offers elastic capacity for compute and storage requirements.
Ordering and provisioning are business/enduser driven, from a service catalogue.
All policy enforcement and management functions are topology-gnostic - and largely independent of typical identifiers in non-cloud archetectures (including IP, Port/vlan/subnet mapping).
In fact, the ability for a workload to exhibit elasticity, practically requires topology agnosticism.
"Flyin' in just a sweet place,
Never been known to fail..."
Great post, but I do disagree with this:
but it's a hell of a lot cheaper, faster and easier than running such a cluster
This is at best only true on a small/medium scale, but ultimately it really depends on your use case and how on top of your AWS bill you are. My contention is that once you get to the point that you're running thousands of large EC2 instances, you'd actually find it cheaper to deploy and manage servers in your own DC (or a colo).
I'm sure the folks at Netflix will disagree with me, but considering the hundreds of folks they have dedicated to tooling and optimizing their AWS deployment, I'd say they're a special case (plus they likely get big discounts given the amount of cheerleading they do for AWS).
True, I should have qualified that statement with "in most cases". In reality, the vast majority of users will never need more than a handful of EC2 instances, and the sort of gains you can get at the small end from using Amazon's pre-built network is amazing. At the upper end, you'll have enough cash available that investing in that sort of talent and infrastructure in-house might be feasible.
Just because you're paranoid doesn't mean there isn't an invisible demon about to eat your face
whooooooosh
(that wasn't a cloud that just whooshed by....)
Here's to the crazy ones
How about you update VMware Converter so it works with the current version of VMware ESXi Hypervisor?
I wasted hours building a system just to find that VMware Converter cannot target VMware ESXi 5.1. Version 5.1 has been out for a month and still VMware Converter Stand-Alone is not compatible with it.
Kriston
It's worked very well for us when we need to run a very paralleled simulation. We run about ten million Monte-Carlo runs per simulation, and being able to tap an arbitrary number of cores for a short time is a whole lot better for us than trying to build it to run for roughly 0.1% of the time and idle the rest of the time.But, we've clearly got an edge case -- looking for edge cases.
App Director is a pretty interesting product. It's rather different from other offerings from VMWare, but it's not a game changer and it's definitely not the future of the company. It's yet another product in the zillions that the company has. VMWare needs to focus. They seem to have lost a sense of direction since the hypervisors have become more of a commodity.
App Director is a nice Flash-based GUI for Chef, which is really the engine underneath doing the heavy lifting. And no, it does not support any other cloud other than Amazon and vCloud Director. It seems to me that VMWare is adding support for Amazon in a couple of their products so they can say "you see?, we do believe in multiple clouds". Should I call it Cloudwashing? There are other more interesting start-ups out there doing similar things like RightScale, Cliqr or GigaSpaces' Cloudify, all of which focus on the applications vs. the VMs. The problem is that I have a hard time believing that VMWare will ultimately abandon the "VM" as the way they see the world.
I manage a cloud with nearly 5000 VMs. In a class today we spun up about 350 on the fly so that engineers could have reference configs available to do their jobs. I'm thinking we'll hit around 6000 VMs by the new year.
Please the first fad were network computers. Same argument was made. The next was intranets. Same argument was made. After that AJAX web 2.0. Same argument was made. Now clouds! ... in reality they are kind of similar. You have a server and you have a client. The truth is I do not have to call it a cloud. An app hosted on a server is just that. Now facebook itself calls itself a cloud and so does amazon. In reality yahoo mail was doing that last century when it was hip to call it a portal.
Yes it is hype with some truth, but is not a revolutionary new thing.
http://saveie6.com/
See what many don't get is IT pros around the would don't like their trade being categorised so easily. What "cloud" is doing is quantifying these layers.
Quantifying the layer simplifies it. If something technical becomes simple then business-heads can understand it (end-to-end) and then what happens is bushiness people do what business people do, that is begin to remove the layer, streamline the layer, outsource the layer, etc. Cloud stops what used to be quite complex thing from being a long laborious costly component to business and quantifies it in a neat little bundle ANNNND when you do that and you can remove / downgrade / retrench the people you need to run it.
IT pros hate it cloud because it gets IT pros fired, makes them less relevant, and less important.
christ on a donkey, but you sound like a fucking salesman.
Architect. ;-)
"Flyin' in just a sweet place,
Never been known to fail..."
Sure, there's nothing it can't do that a properly configured and tuned geographically-distributed redundant cluster of linux boxes couldn't...
It is a properly configured and tuned geographically-distributed redundant cluster of Linux boxes.
When all you have is a hammer, every problem starts to look like a thumb.
$72 a month for mysql 5GB database. That's what I paid last month.