Gartner Article in Network World for reference: Network World-Gartner
I couldn’t agree more with the opening paragraph:
If speedy IT services are important, businesses should be shifting from traditional computing into virtualization in order to build a private cloud that, whether operated by their IT department or with help from a private cloud provider, will give them that edge.
And then interestingly, immediately following that opening paragraph, there is a link to another article talking about a survey that Symantec conducted that shows that many are disappointed in virtualization and cloud computing.
While most seemed largely satisfied with server virtualization, 33% were disappointed in storage virtualization, 26% in desktop/endpoint virtualization, 37% in private storage as a service and 32% in private or hybrid cloud computing. The biggest complaints about hybrid/private cloud computing were in scalability, security and time to provision new resources. In storage virtualization, about a third of IT managers felt they were getting anticipated benefits in operational expense, agility or scalability. And more than a third were disappointed in terms of their expectations about scalability, reduced complexity and efficiency in private storage-as-a-service.
I believe the answer to this issue ultimately comes down to realizing that: “not all cloud solutions are equal even if based on the same underlying virtualization technology” and “you get what you pay for.” The reason I know this is because I have spent most of my career architecting and managing these types of environments. Of course all the media hype about cloud would lead most to believe that it is all the same. If an organization is to move forward with a cloud (or hosted) solutions, they really need to get down to brass tacks. In other words, the customer needs to be asking the following fundamental questions of the service provider:
- How long have you been in business?
- Are you a private or a public company?
- Do you own or lease your data center space?
- Where are your data centers located?
- Is your network platform fully redundant (component-level as well as provider-level)?
- Who are your upstream network providers and what is your overall network capacity?
- Is your server platform fully redundant?
- Is your storage system fully redundant (and to what extent)?
- Who are the manufacturers of the servers/storage systems and what level of support do you have on them?
- Is the overall platform highly available with no single point of failure?
- What criteria is used to prevent over-utilization (and therefore contention) on the platform?
- What is the class or tier of the facility(s) housing the cloud/hosted environment?
- Do you have a tiered electrical system (main line -> UPS -> PDU -> servers)?
- Do you have generators onsite and are they tested regularly?
- How long can you run on generator power?
- How much diesel is onsite and do you have contracts to get more?
- Do the generators have a multistage filtering system so that they don’t have to be taken offline to change the filters?
- Do you have 7x24x365 staffed facilities and technical support?
- How many layers of physical security exists at the facility(s)?
- What data backup services do you offer?
- Do you have the ability to provide site-to-site replication of the virtual systems?
- Can my physical environment co-exist with my virtual environment regardless of where the physical environment is located?
- What organizational certifications do you currently hold (SSAE 16 / PCI / ISO) and how often are you audited?
This is not meant to be an all inclusive list, but as you can see, the questions go far beyond simply asking, “do you offer cloud” and “how much is it?” As a consumer of these services, you will inherit the “quality of service” the service provider is delivering. If there are strengths, you will inherit them. If there are deficiencies, you will inherit them. No system is perfect and no system is absolutely guaranteed, but has the core foundation been architected in a way to provide maximum service uptime and maximum data protection?
“There’s a disconnect today,” said Wolf, noting that a recent forum Gartner held for more than a dozen CIOs overseeing their organizations building private clouds, more than 75% said they were using home-grown management tools for things like hooking into asset management systems and ticketing.
No doubt about it. Every service provider varies in the way it does business. The integration of management functions with back-office functions is essential, but will vary based on the model of services being provided. The integration complexity is directly proportional to the level of billing granularity.
But this private cloud planning will involve updating procurement and change management processes used internally today, as well as figuring out which applications are most suited to be virtualized. Starting off with file and print and simpler applications not considered mission-critical is a good way to start. “some applications can’t be virtualized because they have special hardware requirements,” Wolf said.
I disagree with the mission critical statement. Again, this comes down to architecture. For example, I could run a nice new shiny Bugatti Veyron 16.4 with the tires the manufacturer recommends (and get the performance and agility I expect out of the car). Or I can say, you know what, $30,000 for tires is just too expensive, so I am going to put on some $1000 tires and be done. Well, don’t be surprised if you are driving along and the car doesn’t do what you want it to do. It is the same with cloud platforms, you get out what you put in. When the architecture is built for mission critical applications, the mission-critical applications are the first ones I want on the platform. Ultimately this is where the providers differentiate. I understand that not all customers want or even need the “Bugatti Veyron” equivalent of a cloud solution, but don’t assume that you are getting the Bugatti by default.
Virtualization requires an entirely new way of thinking about networking, said Gartner analyst Mark Fabbi. “Virtualization changes everything, how we look at server I/O, how we deal with branch offices and disaster recovery.”
I agree. It is a new way of thinking, but I would also go as far as to say “it gets administrators thinking about networking.” From my chair, the networking component of the cloud solution has been and will continue to be the more complex discussion with customers. I don’t mean complex as in, understanding the technology. I mean understanding which technology and how best to deploy it given a particular customers’ need. There are many more combinations to consider on the networking side than on the virtualization side. With virtualization, the platform is fairly cookie-cutter (at least that is how we have architected it… making support and deployment much easier). As an integrator of cloud services with many different LAN and WAN options, having the right discussions about the network is critical.
In the end, good communication and face-time are very important elements between the customer and the service provider. Many problems today can’t be solved by simply swiping a credit card and ordering a ‘certain number this’ and a ‘certain number of that’. Is there a need for this type of service, yes. Do I think that enterprises can run their organizations’ critical applications on these types of platforms successfully, no. There must be collaboration up front to get the solution right in order to take full advantage of everything virtualization has to offer (in the cloud).
My Recommendations: Perform good due diligence, dig deep into the service providers’ architecture to confirm it aligns well with your operational expectations, and keep in mind, “There’s no free lunch”.