Cisco Blogs
Share

The Long Road to the Cloud – Changes in Application Deployment Criteria


September 3, 2015 - 2 Comments

In today’s world as more and more customers prepare to take advantage of cloud technologies, they are finding that private cloud and colocation services are essential options in their journey to the cloud.

Harrington_DanWe are lucky to have Dan Harrington, as a guest blogger. Dan is a Research Director covering Datacenter trends at 451 Research. His primary focus is managing 451’s Voice of the Enterprise: Datacenters study which surveys thousands of enterprises a year about their datacenter strategies.

Out of the insights of his surveys, Dan has agreed to share:

  • The most important criteria are when determining whether to deploy in your own datacenter, at a colocation provider or in the cloud.
  • Where IT organizations are deploying their applications, today and in the future.
  • How security is often the most important criteria when determining deployment location.

If you believe what you hear from the mainstream media, investment community and tech press, you may come to the conclusion that every application is being deployed to the cloud or an off premise colocation datacenter. And that the very idea of deploying in a company owned datacenter went out of fashion long ago. After all, Amazon Web Services is currently pulling in $6bn annually, which is quite impressive – regardless of the fact that the entire IT industry is worth well over $1 trillion a year. However, if you look under the covers you will find that IT organizations still care very much about attributes that don’t necessarily always lend themselves well to an off-premise deployment. Learn more about which vendors are leading the market in IaaS and on-premises cloud platforms.

VotE_DC_Q2_2015_AppCriteria-03 (8.24.15)

N=416 Source: 451 Research Voice of the Enterprise: Datacenters, Q2 2015

A large (>1,000 Employees) Public sector organization weighed in last quarter about what he considers when deploying a new version of Oracle:

“The most recent major application [workload implemented] is more of an upgrade to Oracle 12… There weren’t really any alternatives [about where to deploy it]. It was here or our colocation facility… Keeping it on [premise] is important, but I think one of the main issues would be just network reliability between here and the colo… We’ve got staff here that are ready and able to deal with any kind of network or server issue. But it would take us an hour or so to get out to the colo site.”

Ultimately the application has to be reliable and perform well. It simply has to work. It may seem obvious but without those table stakes criteria then a deployment location, whether colocation, cloud or on-premises, will likely not be considered. Just as important as whether or not it will actually run well, is how secure it is. This is top of mind for almost all of our respondents due to all the high profile corporate IT failures such as Sony, Target, Home Depot or even the US Government. Utility companies, financials, healthcare and government all rated security as their highest priority. Cost and time to deployment, often touted as benefits to cloud computing, were cited as important but not necessarily critical for choosing a deployment location. Learn more about what end-users say their leading security drivers are.

Regardless of how you look at it, enterprises are leveraging more colocation and cloud resources, which is reducing their reliance on their own facilities. However, this is not happening at the drastic rate that some vendors may lead you to be­lieve.

VotE_DC_Q2_2015_REV2 (8.24.15)

N=475 Source: 451 Research, Voice of the Enterprise: Datacenters, Q2 2015

Organization’s pace and strategy for adopting cloud vary widely and are influenced by numerous factors.

One large (>1,000 Employee) financial organization detailed their thought process:

“[If we run out of capacity we would use] a combination of the colo, SaaS, PaaS. . . We’re designing our environments and our strategies to include extending our infrastructure out into the cloud, so we’re spending a lot of time not doing capacity management and not figuring out how I would build the next data center. I’m trying to come up with the methodology and the strategy and process to say, “I want to be able to extend my current infrastructure into the cloud. And, how do I do that in a secure manner? How do I keep my penetration test – IDS, IPS, the compliance requirements, and the firewall and the segregation of those systems? How do I move that out into somebody else’s data center?” Instead of looking at data centers, we’re looking at the methodology process and standards, and strategy to help us expand into the cloud and that’s probably the route that we would go.”

Within three years, enterprises expect that one in five applications will be deployed at a cloud service provider, and one in seven at a colocation provider. Clearly, cloud and colocation providers are maturing at a rapid pace. At the same time, organizations have significant investment already poured into numerous datacenters (watch Dan’s recent webinar, Enterprise Datacenter Investment: Facility Modernization and Capacity Allocation). Enterprises will continue to utilize existing sites to ensure that their applications are protected while also looking to leverage colocation and cloud services providers to enable new capabilities. Anyone who expects a wholesale cloud shift in the short term, likely has their head either in the clouds… or the sand.

To learn more:

Get access to the 451 report Voice of the Enterprise: Datacenters study
Cisco ranks highest for promise among hardware-centric vendors
Follow Dan or me for more updates on Twitter.

In an effort to keep conversations fresh, Cisco Blogs closes comments after 60 days. Please visit the Cisco Blogs hub page for the latest content.

2 Comments

  1. Added note: sorry for the Block O'Text comment formatting above. I thought my draft would retain formatting and, nope, it didn't. Ha ha, but since it probably didn't get read, apparently not an issue.

  2. “We’ve got staff here that are ready and able to deal with any kind of network or server issue. But it would take us an hour or so to get out to the colo site.” Why would a site need to send its own techs to a remote site? What if the site was across the country? What if it was a CDN network? That’s the whole purpose of including remote hands services when one positions computing resources other than locally. Anyway, the application, whether local or remote, needs to be architected for resilience in the case of server failure. After all, servers can fail if they’re local and, presumably, the app needs to keep providing service. Similarly for network failure, one architects for network degradation or failure. “[If we run out of capacity we would use] a combination of the colo, SaaS, PaaS. . . We’re designing our environments and our strategies to include extending our infrastructure out into the cloud, so we’re spending a lot of time not doing capacity management and not figuring out how I would build the next data center.” What does it mean to “extend infrastructure to the cloud?” Sounds as if “cloud” just means traditional co-location in this context. Where is the data that the applications are reading and writing? I’m surprised that data affinity isn’t one of the top factors considered when considering using remote compute services, whether classic co-lo or true IaaS, elastic, self-service “cloud” compute/store/network.