I wanted to take some time and make sure that I clarified my position on Openstack. Rackspace and NASA announced this open source project back in July of 2010. Openstack is a cloud manager for private or public clouds. Openstack positions itself as an alternative to vCloud and through 3rd hosting provider party an (potentially) open standards alternative to Amazon’s AWS.
So, I was quoted in a GigaOM article that I felt Openstack and by proxy Rackspace may be little late to the cloud manager game. The jest of some of the posts was that this is still a relatively new market and there is plenty of time for new competition. I don’t totally agree with the full statement. I do believe that we are early in the market but I don’t agree that there’s plenty of time for competition.
My primary point – If a customer wanted to roll a cloud today what are their options? Rackspace has offered Cloud based services for a long time. One of their challenges has always been their lack of API into their IaaS based solutions. That’s one of the reasons why they decided to sponsor Openstack. And they now offer their Openstack Cloud in limited availability. One of the main strengths of AWS includes the API’s. There’s obviously some great scalable applications built on the Amazon cloud which has been in production since 2002. Zynga is a great use case for AWS and use of the AWS API to build a hybrid cloud. So, past performance is a critical consideration to overcome for enterprise customers considering Openstack.
Building a cloud manager is hard work. Cloud managers are extremely complex orchestration systems that are part data center automation, CRM and accounting systems. I guess you can think of them as ERP for the cloud. So, I’m not saying that the Openstack development crew is taking too long to deliver their solution; I’m saying the project may have been kicked off too late.
When I talk to customers today about public clouds, I’m constantly asked about AWS in addition to other solutions I put before them. In my non-technical customers AWS already has a huge part of their mind share when they think public cloud. When I talk to the today about private cloud strategies they and we talk Openstack they ask me who has it in production today? How does it extend to hybrid models today? I believe Openstack will have a great story to tell some months from now. But that’s the challenge. It’s not vaporware the code is out there to download and install and jump start a cloud development project. But if you want a production ready solution now…. vCloud, Abiquo, NetIQ, Dell, Eculyptus are all out there waiting for you. I don’t believe AWS is too late for private cloud but public cloud is looking like a one horse race today.
So, today what are end user options for cloud? What companies come directly to mind when asked the question? Is it too late for the Openstack consortium?
So, I normally teach this class as a two hour session and I had a heck of a time cutting out material and getting it down to something digestible for Internet (and slide deck) viewing. It’s a quick introduction to cloud managers and a continuation of my introductory cloud computing course.
I just listened to a talk from Berkeley professor Scott Shenker yesterday on youtube http://www.youtube.com/watch?v=WVs7Pc99S7w that gave an excellent breakdown of SDN and he spoke of the need of the Network Operating System before SDN’s can become a reality.
When I think about it, I’m rather amazed that we haven’t created an abstraction for the network. His talk speaks about how relatively easily we’ve done this at layer 2 but how difficult it is to do at higher layers due to the non-modular design of the network stack. Applications shouldn’t be making calls to the network address but rather to the network service.
Interesting stuff. OpenFlow is a step in the right direction to creating the “BIOS” that we need. I’m especially happy that Google is at the bleeding edge of this in a production network.
Today my employer hosted an internal conference and Microsoft just happened to be there which I wasn’t expecting. Specifically, they were there showing off Windows 8 running on a Samsung Series 7 11.6” tablet. I had ran the Developer Preview of Windows 8 on my older HP tablet which was an OK experience but the hardware was not optimized for Windows 8.
The Series 7 is designed for Windows 7 but I have to say that the early experience was impressive. The hardware is not as portable as my iPad but I consider this tablet to be a different use case than the iPad. MS had a wireless keyboard and the very stylish and sleek docking port for the tablet. I quickly imagined this being my portable solution. I’m actually tempted to get the existing setup running Windows 7. My current laptop is more of a desktop replacement machine and every time I travel I wish I had something along the lines of an ultrabook but the flexibility of a tablet.
My early experience with Windows 8 running on this tablet was pretty nice. The MS rep pulled out his phone and compared the Metro interface on the Windows 8 Phone to the Metro Interface running on the tablet. He spoke to a couple of use cases with factory workers and integration with ERP applications. It is a compelling story from an enterprise support and application development perspective.
I’m looking forward to getting a couple of these in the lab and really stretching it out a bit. I’m really interested in the Intel platform and seeing how end users actually take to these devices and what type of real world battery life they get.
My wife asked me the other day how does she log into her iCloud and get some data? This is the problem with Apple’s cloud strategy – My wife doesn’t understand it. My wife is a typical user – not a geek or even a power user. She uses her computing devices as a tool unlike most readers of my blog. She is the typical person that would walk in the Apple Store and purchase an iPad, iPhone and MacBook Pro (she doesn’t get the appeal of the Air). Just casually she expected to sync her documents, music and pictures to her iCloud and access it from any of her Apple branded devices and access the data via an iCloud website. She perfectly understood if she couldn’t access the data from her PC. After all it is the “iCloud”.
When I tried to explain to her that it didn’t work that way, she just had this blank look on her face like “Isn’t this an Apple solution?” When I showed her the first version Photo Stream and how you can sync pictures across iOS devices she initially was ecstatic until she found out you couldn’t delete any of the pictures. Her question was “Why doesn’t this work like that Dropbox program you installed for me? It just works.”
She appreciates iTunes match but doesn’t understand why should doesn’t have that functionality across all of her data. I couldn’t have captured Apple’s challenges with cloud any better.
Microsoft’s VM-Limited campaign has pitted VMware’s cloud strategy as just more virtualization masked as a Cloud solution. Microsoft hits on some of the pain points of VMware’s customers including the change in licensing model which the faux VMLimited sales person explains it as “The more you use the more you pay” model. From the density you look to get from a cloud solution Microsoft may have a valid point. The new model for vSphere 5’s licensing does at face value makes you believe you may pay more if you offer really dense nodes. I’ve looked at the licensing model and you really do need to have a specific use case to exceed the current memory limits. I could see how cloud provider who look to get the most density out of their physical hardware would be at risk.
Another knock on VMWare is one of my main grips with vCloud which is its inability to support any other Hypervisor other than ESXi. I interested in helping organizations build private and hybrid clouds and it is very easy to make a case for creating a private cloud based on more than one hypervisor. I could see ESXi being used to host critical non-stop workloads and KVM, XEN or Hyper-V being used to host less critical workloads (not that they aren’t capable). It seems to reason that a mixed hypervisor environment can save organizations deploying private clouds money. I think VMware would answer by saying if you need to service those use cases then you could always use ESXi free. However, you can’t manage ESXi free with vCloud Director.
The third point I get from the video is that VMWare doesn’t give your insight into your application. I’m not too sure about this point. Even if you just consider virtualization there are plenty of tools that help you with application performance. I think this was more a reference to Microsoft’s Azure cloud solution which is more of a PaaS. VMware does have its CloudFoundry project platform which addresses PaaS. It is fair to say that Azure is much more mature than CloudFoundry at this point.
Whether the campaign is accurate or not it is one of the funniest campaigns Microsoft has put on. I do believe you have to be a virtualization geek to get the references and may fly over the head of some decision makers. Too the cloud indeed Microsoft
I’ve been in enough cloud engagements to realize that the Cloud is not (just) about saving money. Sure, you can reduce start-up costs and pay “On-Demand” for resources. If you have 1001 e-mail users you can subscribe to 1001 instances of Google Apps or Office 365 and pay for exactly what you use. Does that mean you’ll save money over deploying Exchange or some other groupware? Maybe. The focus shouldn’t be on just savings rather the functionality gained by the cloud.
In this Cisco video Forester talks about some of the drivers for engaging a cloud strategy and the business stakeholders making the case for cloud. It’s not always cost savings or IT. In most instances that I’ve been involved in it’s about capability. Why does Netflix stay with a cloud for distribution? I’m sure at this point they could build out their own data centers and probably save money over AWS.
For each business, it will be a different driver. For Netflix, I believe the driver is the scalability of AWS. And not just AWS, if projects like OpenStack/CloudStack take off then organizations leveraging the cloud will have a much larger pool of IaaS providers to choose from if they aren’t already leveraging other cloud providers. You can’t easily build that type of redundancy. Even if Netflix chooses to bring their compute in-house, I’m sure they’d benefit from some type of hybrid public/private cloud model.
We have Netflix on one side of the coin and small enterprises on the other. Why engage Google for a Google Apps deployment? Creating an LAN with servers, remote access, office applications, e-mail and collaboration tools is an old trick at this point. This has become a commodity solution where any jack of all trades can get your business up in running. What they can’t do is give you the seamless integration of services and support that a Google Apps or Microsoft 365 can give you. You can also easily share data between your organization and your business partners without worrying about firewalls and user accounts.
A great example is that I use WordPress.com not because it’s less expensive than hosting my own WordPress blog on AWS or any number of hosting providers. I’m obviously capable of hosting a virtual server running WordPress. I choose to run on WordPress.com because I’d rather spend my time doing research and writing vs. patching and maintaining.
Of course, there are considerations to have when it comes to security and features but these issues are slowly but surely being resolved.