One of the definitions of the area “cloud” in Merriam-Webster’s dictionary is “something that obscures or blemishes”.

This definition often seems fitting for today’s computing cloud. I actually have seen quite a few universities move services to the cloud only to reverse course and move them back to campus as a consequence of “blemishes” in security, reliability, service, cost, integration issues and/or compliance concerns. In general, at Southern Illinois University (SIU) we’re currently moving faculty and staff email back to campus. And it’s not just higher education. A Gartner analyst recently predicted that by 2014, 30% of enterprises using software as a service will move back to on premises systems.

[ How do you get more robust online services? Read 7 Habits Of Successful DevOps. ]

Personally, i’ve been evaluating cloud services for higher education for the previous couple of years and feature rarely found them a fit for my organization, even though yearly cloud computing is a top IT issue for higher education. It was rated the #3 issue in 2013 in a up to date Educause report. The criteria that make cloud services less attractive to our industry include:

1. Cost

Despite the rhetoric, cloud services are usually way more expensive than hosting an on-premises system. i feel that that’s very true for public higher education because that we receive tremendous discounting on most technology and traditionally pay much lower salaries than the personal sector. This ends up in an incredibly low total cost of ownership (TCO) for our self-hosted services.

2. Reliability

At SIU, we outsourced our learning management system at a miles higher cost than the former on-campus system. One of several reasons that we paid this extra cost was because we expected higher availability from the cloud-based system. Unfortunately, we’ve experienced precisely the opposite. After two years at the new system, our downtime was consistently and significantly higher than we experienced with on-premises solution. There were times that the system have been down for a whole weekend. Additionally, I even have other reliability concerns with the outsourcing experience:

— By outsourcing systems we lose the everyday control. Once we do have an outage (or other issue) we’re solely reliant at the vendor. With an on-premises system, we’d commit all necessary resources to mend an issue expediently. With a cloud system, we’re helpless.

— By outsourcing to the cloud, we create multiple additional points of failure. We’re still reliant on our own infrastructure, but we’re also reliant at the infrastructure of the seller, their Internet merchant, our Internet corporation and all points in between.

Unfortunately, our experience with loss of reliability isn’t unique. Even the well-known public cloud providers have suffered significant outages during the last couple of years.

3. Security

As a tremendous research university, we now have a considerable number of legal obligations in regards to the protection and preservation of our data. Areas of compliance include FERPA, HIPAA, PCI, FOIA, litigation holds, electronic discovery and more. We recently have seen quite a few cloud providers which have severe security breaches (Amazon, Microsoft, Apple, Google, Sony, simply to name a couple of). Their very size makes them a large target.

Another security concern is that the majority cloud providers have data centers around the globe, that are subject to the jurisdiction of local governments. The former university I worked for decided against outsourcing student email to Google for this very reason.

4. Bandwidth

Being located in Southern Illinois has a number of benefits (beautiful scenery, plenty of wineries, and southern hospitality), but Internet bandwidth isn’t one in all them. We have now a really limited amount of fiber infrastructure in our area, which ends up in both a limited amount of obtainable bandwidth and a miles higher cost for that bandwidth. By moving applications to the cloud we consume additional bandwidth at an incredibly real, significant and measurable cost that impacts the TCO.

I’m in no way saying that universities shouldn’t consider moving applications to the cloud. As a matter of fact, SIU has very successfully implemented a handful of cloud services and applications. However, I do offer the subsequent unsolicited advice to my colleagues in higher education:

Don’t …

— Move to the cloud because it’s easier to procure budget for recurring expenses than it’s miles for a capital expense and permanent staff positions. Although this may must be done out of necessity (I’m guilty of it), it may result in an overall increased cost to the university.

— Move to the cloud simply for the expediency of shorter project timelines. It is tempting but could have a protracted-term negative impact to your university.

— Move to the cloud without an ironclad service level agreement guaranteeing uptime with penalties for downtime.

— Decide to long-term contracts for cloud services.

Do …

— Investigate private cloud and community cloud offerings (corresponding to Internet 2’s Net+ services).

— Evaluate all your options, using a way that considers TCO and return on investment together with institutional risks and business continuity.

— Investigate costs and complexities of integrating the cloud service into your existing environment.

Perhaps if we’re deliberate and judicious in our use of the cloud we will be able to avoid seeing it fit one other cloud definition from Merriam-Webster: “Something that has a gloomy, lowering, or threatening aspect.”

The crucible of cloud, big data and distributed computing is hell on systems. Will application performance management settle down complexity — or simply add fuel to the hearth? Also within the new, all-digital APM Under Fire special issue of InformationWeek:Cloud industry heavyweights discuss the professionals and cons of OpenStack support for Amazon APIs. (Free registration required.)