Transforming the Datacenter


Agnosticism rules the cloud

Agnosticism rules the cloud

Summary: At some point in every IT manager’s career, someone somewhere above them in the organization is going to make a cloud decision based on a misconception. And these managers are going to blow it (and probably millions of dollars)—big time.

SHARE:
TOPICS: Data Centers
6

At some point in every IT manager’s career, someone somewhere above them in the organization is going to make a cloud decision based on a misconception. And these managers are going to blow it (and probably millions of dollars)—big time.

My advice for anyone who ends up in that position? Heed the learning experience because it will happen to you again. Meantime, it may be possible to head off some awkward moments by understanding some of the issues. For example, look at public vs. private clouds.

The notion of private or “on-premises” cloud has sparked heated debates in the technology community. Well-respected experts contend there is no such thing as a private cloud. They argue that on-premises datacenters lack infinite elasticity, which is an essential element of a cloud service. Some critics also object to the absence of flexible billing or the fact that services do not necessarily transit the public Internet. If we take some of the prevalent definitions of cloud computing literally, it’s not hard to understand their argument. Yet both have advantages. The biggest difference between deployments in the datacenter and deployments in the cloud is designated hardware.

On-premises clouds operate essentially the same way as public clouds; we just use different terminology to describe them. Enterprise business units obtain their services from the IT organization in much the same way as public or “off-premises” cloud customers subscribe to their services from external suppliers. What’s the difference? In one case, the consumer and provider belong to the same legal entity, while in the other they do not. On-premises clouds achieve economies of scale by pooling internal resources. They may use the same technologies for virtualization and automation that are found in off-premises or public clouds.

Private and public clouds, both on and off premises, also face many of the same administrative and infrastructure challenges. Just as public clouds need to manage their accounts, private clouds need to be able to allocate costs. Private clouds also need to logically separate co-located tenants and applications, even though they are internal. And frequently, changing workloads and projects are only feasible if the infrastructure is dynamic and highly automated.

The bottom line: good cloud architecture needs to be agnostic with respect to delivery point and should work equally well for private as well as public, on-premises and off. The question isn’t so much whether to refer to a private cloud as cloud, but whether it is meaningful to make the distinction. Corporate datacenters can be turned into on-premises clouds with all the advantages of performance and scalability. My take? A cloud is a cloud, and clouds used in many different ways benefit from common approaches to infrastructure management. And there are sound infrastructural reasons why companies put some IT resources into on-premises clouds and others off-premises.

This content was published by CBSi B2B's custom content group, which powers this Microsoft sponsored blog. 

Topic: Data Centers

John Rhoton

About John Rhoton

John Rhoton is a contributor to CBS Interactive's custom content group, which powers this Microsoft sponsored blog. He is a technology strategist who specializes in consulting to global enterprise customers with a focus on cloud computing.His tenure in the IT industry spans over twenty-five years at major technology companies, defining and implementing business strategy. He has recently led corporate technical strategy development, business development, and adoption of cloud services, datacenter transformation, mobility, security and next-generation networking, while also driving key corporate knowledge management and community-building programs.John is the author of six books.

John Rhoton's views are his alone and do not necessarily represent those of Microsoft or CBSi.

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

6 comments
Log in or register to join the discussion
  • Biggest advantage...

    " They argue that on-premises datacenters lack infinite elasticity, which is an essential element of a cloud service. Some critics also object to the absence of flexible billing or the fact that services do not necessarily transit the public Internet."

    Surely, not transiting the public Internet is the biggest advantage of such "private" clouds? If the private cloud is not public facing, there is less of an attack surface. Access only through VPN and having it locked up behind a decent security infrastructure is a good thing, if the data is a of a sensitive nature.

    Expandability is also more limited within a company. It will need to expand as the company grows and the data pool grows, but it doesn't need to be ifinitely expandable, because there is a very finite number of users. As long as the systems are designed with expandability in mind, throwing new hardware at it, as it is needed, is more economical than having thousands of redundant servers kicking their heels.

    Both types of cloud have their advantages and disadvantages, which one you use should depend on your needs and how sensitive the data is.
    wright_is
  • Each model has its advantages -- but flexibility is always critical

    Hi wright_is --- thanks for your comment!

    You make some good points with regard to additional distinctions between private and public clouds. I guess you could argue as to which is more important, which would vary from one company to the next.

    Not transiting the Internet is important for some. Organizations can usually secure the connect cryptographically to prevent data leakage while the data is in transit, but it still adds to the potential exposure.

    Expandability will also vary from company to company. Some large corporations may have more potential for expansion than small service providers, but clearly the tendency is for public providers to have more resources than private datacenters.

    In any case, as you point out, there are advantages and disadvantages to both, which I hope is also clear from the original post. My perspective is that a well-designed architecture can accommodate either delivery, so that the company is free to choose to option that best meets their needs --- and to change it as their needs evolve.

    John (@johnrhoton)
    johnrhoton
  • A cloud is a cloud?

    Good! Now, what exactly is a cloud?
    dratman
  • What a crock

    I remember when the "on-premises" cloud was just called the LAN. The quality of your LAN was directly tied to its architecture and the know-how of the people building and maintaining it.

    The LAN was expensive - both in direct cost of equipment and the cost of hiring staff to service it.

    The Cloud came along and promised a reduction in the need for both equipment and staffing . . at the obvious cost of security. But it offered you the ability to outsource all your architecture and a majority of your IT staffing at either a performance boost or a drop in cost.

    They promised you speed, reliability, and an ability to upgrade your service within moments. Need more storage? Done. Need more servers? Done. Need more bandwidth? Done. All for a service fee with little to no waiting.

    A private cloud is a fancy way of saying, the return of the LAN.
    wildeep
    • wildeep speaks truth

      "The cloud" was just another buzzword for techniques for handling virtual storage. The third party has flexibility on what servers to put it on, and it can use system software to handle these multi-system databases. I really don't like that word, "cloud".

      Google told one company I know about it could not promise that the "cloud" data would stay inside US territorial boundaries, and lost the account. What kind of security is that, especially with what we now know about the spying. Consider that for that to work, what we now know publicly, for it to work, somebody has to have some back doors. How do you know the data Google promised to protect for you, how can you know whether they are under a gag order about anything at all?

      At least if it stays under your roof your competitors can't get it. Think they can't? Well, think again about one IRS low-totem-pole guy that gave ProPublica a confidential list of donors that not even the IRS is supposed to have. That's political, but why don't you add the element of money?

      Oh, won't happen in your compliance-driven company? Never heard of any rogue employees, eh? Can't happen to you? Why take the chance?
      infotechadviser
  • "Infinite"? Let's talk sense

    For every "infinitely" elastic cloud, there is a botmeister out there who can overwhelm it. Besides, exercising infinite elasticity, whether in-house or outsourced, would cost an infinite amount of money, which does not exist.
    cowan@...