Private Vs. Public - It’s About The Services

So you’ve chosen to implement a private cloud, and you based that decision on sound rationalizations backed by thorough research--reasons like cost, service portability, legacy infrastructure investment, security and compliance. Heck, maybe you just made the decision based on an overreaction to recent cloud outage news from major providers. Either way, private cloud it is, and your decision on cloud type is done, right? Wrong. There’s a lot more to think about, and if you choose priv

Joe Onisick

June 15, 2011

4 Min Read
NetworkComputing logo in a gray background | NetworkComputing

So you’ve chosen to implement a private cloud, and you based that decision on sound rationalizations backed by thorough research--reasons like cost, service portability, legacy infrastructure investment, security and compliance. Heck, maybe you just made the decision based on an overreaction to recent cloud outage news from major providers. Either way, private cloud it is, and your decision on cloud type is done, right? Wrong. There’s a lot more to think about, and if you choose private cloud with an all-or-nothing strategy, you’re likely missing out.

Just because the bulk of your business applications and services are slated to migrate from legacy architectures to fluffy cloud architectures within the four walls of your corporate-owned data centers doesn’t mean they all should. Many services and applications you run may still gain flexibility, scalability or cost savings by moving to the public cloud. Additionally, your business/mission as a whole may gain agility and competitive advantage by maintaining a partial foothold in the cloud. This second part is key because cloud isn’t just about cost savings. In fact, in some cases it may involve extra costs. Cloud is about returning IT to its rightful place as a business enabler rather than the current spot of "cost center."

There are many options when incorporating private and public cloud strategies into a cohesive service delivery model. One commonly discussed model is the hybrid cloud. Hybrid clouds are a mix of two or more of the National Institute of Standards and Technology (NIST) cloud types, usually discussed as public and private hybrids. Examples of hybrid clouds may be:

  • Public cloud-based disaster recovery systems backing up private cloud services

  • Hosted infrastructure as a service (IaaS) used as a business continuance platform

  • "Cloud bursting" unpredictable capacity to a public cloud system ( For more thoughts on this, seemy post: The Reality of Cloud Bursting.)

    The other option (although not mutually exclusive) is to assess individual applications or services to determine the best location--private or public. Many applications lend themselves well to public cloud models. Email is a perfect example because:

  • It’s not latency sensitive.

  • It experiences large variances in compute requirements throughout the day/week/month/year.

  • The content will spend the majority of its time on non-internal systems anyway.

  • It can be difficult and expensive to manage correctly internally and may be handled better with scale.I’ve experienced a half-day corporatewide email outage based on a single misguided user believing that Bill Gates would pay her $200 for each person she forwarded a chain mail to. She added every email address and alias in the corporate directory to the forward, including conference rooms and scheduled resources. The effort of processing single copy forwarding among the groups, auto replies, and so on brought the system to its knees. The half dozen Reply All corrections/reprimands from others didn’t help. This type of outage shouldn’t happen in a public cloud hosted email system because the hugeburst in traffic from the affected corporation would be offset by the underlying infrastructure designed to support hundreds of like corporations.

    Many other applications are similarly suited to a public cloud environment. And very few organizations, both government and commercial, are incapable of mixing the two approaches, as long as applications are looked at individually. Some hypothetical examples:

  • A government intelligence agency decides its data is too sensitive for public cloud resources and chooses a private cloud model. These agencies still run public services, websites, and portals that contain non-sensitive data and are accessible in some form to the general public. These services are perfect candidates for public cloud because the cloud frees up internal IT resources to focus on delivery and security of the sensitive services rather than building a series of separate security enclaves internally.

  • A large enterprise analyzes IT costs in a public model vs. private model and determines that at its current IT efficiency, private cloud is more cost effective than a public model. (See the Wikibon post on this.) In this case, the bulk of the IT services will remain in house, but services like email, instant messaging, Web hosting or others may be more cost effective or flexible offsite.

  • A financial institution insists on private cloud for compliance concerns. Compliance covers the personal data of the institutions clients but not all data the institution touches. Services like CRM, email, portions of web hosting, and many others may be excellent candidates for public cloud while maintaining compliance sensitive data in-house.

    Cloud computing offers the ability to enable IT to drive business agility. That ability is amplified if each service is looked at with the following two questions: What is it I am trying to do, and what is the best way to accomplish that?

Read more about:

2011

About the Author

SUBSCRIBE TO OUR NEWSLETTER
Stay informed! Sign up to get expert advice and insight delivered direct to your inbox

You May Also Like


More Insights