A risky cloud approach?
Apologies to Shakespeare. He did not know about cloud computing, the technologies that deliver software and services from the Internet. He did know about the common sense of putting on protective storm gear. What would he think about today’s cloud approach? I think Billy would pull on rubber boots and a sturdy yellow slicker. The computer clouds are unsettling.
First, how do I spot the person responsible for in-house, on-premises computing systems, enterprise software, and the bells and whistles like routers and firewall that are mandatory?
I just look for the person who looks tired and hassled. Those characteristics fit some of the professionals involved in information technology (IT). You may have a different point of view, but I think keeping an in-house server complex online and stable is one of the toughest jobs in an organization.
Hardware is getting better, but when a disk drive fails, the gizmos have to be replaced. Compared to software problems, hardware is a minor glitch. Software is the migraine maker. Applications and their interactions have become complex. Mysterious interactions among systems create "What do we do now?" situations.
Chief financial officers know that costs for information systems and infrastructure are tough to control. The reason is that when an in-house system goes down, no one knows how long it will take to remediate the problem or how much the repair will cost. Unpredictable and unknown costs consume an IT budget with alarming velocity.
When an important proposal must be sent to a key customer or a legal document is due before a court-imposed deadline, costs become a secondary consideration. I have heard many times this statement, "Just get the system back online." Even a minor glitch with a printer hooked into an enterprise publishing system can wreck havoc with budgets. As good as an in-house engineering team might be, some problems require specialized skills. A shallow hiring pool translates to expensive consulting engineering support. In fact, in-house IT professionals may not be able to troubleshoot the problem accurately.
In response to those on-premises cost issues, cloud-based services have sprouted like mushrooms. The cloud computing model makes sense. First, the cloud service vendor can distribute the expense of specialized engineering talent across a number of customers. Experts are available to deal with problems. Second, software as a service (SaaS) can be purchased with a service level agreement (SLA) tailored to the needs of the licensee. On-premises software may not be available with an SLA-approach to uptime, performance or other performance guarantees.
If the promise of cloud computing is accurate, an organization can reduce certain hardware costs, some license fees and maybe trim headcount. Instead of a baffled grin from the in-house expert on SharePoint, the CFO gets the teeth from a service level agreement with a cloud solution vendor. The idea of having "one throat to choke" is enough to whet the appetite of the most cost hungry accountant.
The argument for cloud computing for organizations is a compelling one. But organizations are conservative beasts. Furthermore, the agitation rippling across the Web logs and Internet news services may be increasing concern among some senior managers about the reliability, stability and security of cloud services.
Consider what’s happened in the first half of 2008.
On Aug. 11, an estimated 20 million Gmail users lost access, causing a huge furor. News of the outage spread like wildfire, prompting both outraged and distressed blog posts and more informational details (techcrunch.com/2008/08/11/systemwide-gmail-outage).
Amazon has suffered from multiple outages despite positive self-issued publicity. A major S3 cloud-based storage program dump occurred in early 2008 (http://gigaom.com/2008/02/15/amazon-s3-service-goes-down). Then Amazon’s core service suffered service outages on June 6 and again on June 9. Still another outage occurred in July. The information about the problem has been slow in coming. Regardless of the cause of a particular outage, those recent problems suggest that Amazon’s engineering has some details to which the e-commerce site must attend.
Add Yahoo to the list. There were outages in June 2006 (http://news.zdnet.co.uk/internet/0,1000000097,39276706,00.htm); February 2007 (http://gigaom.com/2007/02/12/yahoo-mail-down); July 2007 (Yahoo outage caused by Level3 BCP issue); and, most recently, August (http://solsie.com/2008/08/yahoo-mail-is-suffering-outage).
Even more distressing for users than service interruption is losing their data. One of the reasons why Apple’s newly launched MobileMe service has been nicknamed "MobileMess" by some is because users have handed over $100 to Apple only to find that every e-mail they’ve ever sent or received has disappeared forever (http://pogue.blogs.nytimes.com/2008/07/24/apples-mobilemess).
Not even enterprise software-as-a-service leader Salesforce.com is bug-free. A February 11 outage left
customers without mission-critical data for about six hours (itbusinessedge.com/blogs/hdw/?p=1639)—another prime example of paying customers having trouble.
Not surprisingly, "little guys" like WestHost have experienced problems too, and claims of 99.99 percent uptime do little to reassure customers like me. On June 19, utility problems caused a 40-minute outage.
In August, my own Web log was unavailable due to "unexpected issues." My operation is a small fish in a big sea, but I can tell you I don’t have much faith in outfits that can’t keep their cloud service from raining on my parade.
Taking all this into account, here’s my thinking:
1. Cloud computing won’t work if it is not reliable.
Despite the keening from the masses wanting their services back, it won’t take too much for people to start looking for alternatives. In February, cnn.com posted an article (http://bigtech.blogs.fortune.cnn.com/2008/02/29/cloud-computings-reliability-gap) that probably (hopefully) put fear into RIM (Blackberry), Amazon S3, Google /YouTube and Microsoft Hotmail. At stake: business models for those trying to ditch self-supported software, goodwill toward specific companies and billions of dollars of investments in cloud services.
"Availability is essential in cloud computing," says Thomas Vander Wal, founder of the IT consultancy InfoCloud Solutions in the following article: technologyreview.com/read_article.aspx?ch=specialsections&sc=storage&id=21133&a=&a=f&a=f&a=f. "If constant access to information and objects is a requirement, then cloud computing may not be a viable option without alternate solutions."