"If you can bring me an architecture that says, 'Here's your tier of virtual services you can manage,' and you can say there's virtual as well as physical security all up and down the stack, now you have my attention," said Michael Baum, chief corporate and business development officer and co-founder of IT log search vendor Splunk. "But until you get to that stage, you're not going to see a lot of use for enterprises for cloud computing."
Many businesses remain wary of cloud computing, worrying about quality of service, uptime, security, and other manageability issues like authentication. "I don't just want to manage the app, I want to manage the service," said Alistair Croll, a BitCurrent analyst. "I want to manage the authentication service; I want to manage the network service."
One of the main things many cloud computing vendors lack today is a sense of visibility into the infrastructure running the cloud, Baum said. "The problem with Google App Engine is, I can't even see if an application is slow until a user calls me," he said. "There's zero transparency. What we've learned from decades of enterprise app management is I need to have my own transparency into the application and the other physical and logical tiers my application is running on."
However, such transparency and the tools needed to adequately manage cloud services to the breadth and depth of on-premises enterprise application management may be years away. "With time, you will actually get a management solution that you can go to, but with time," said Vipal Sharma, co-founder and VP of engineering for virtualization management vendor Fortisphere. "It took 10 years for IP networks to get to that point."
One of the fundamental aspects of the cloud that makes it hard to manage is multitenancy. Since so many other companies are running virtual instances of applications completely independent of one another, customers may have little ability to control infrastructure issues and have to leave some management up to their vendors.
"What do you do if you have I/O scheduling issues because some guy on another part of the server is running some runaway database?" asked Reuven Cohen, co-founder and CTO of private cloud company Enomalism. A recent Amazon EC2 outage, for example, was tied to a piece of inefficient code that took down the whole system.
To help understand how other companies large and small are approaching cloud computing, InformationWeek has published an independent report on the subject. Download the report here (registration required).