Managing the Cloud

  by Adam Kerrison | 24 Feb 2009 | Permalink |

Cloud computing has come up in conversation a lot recently. It’s clear that IT is moving toward a service-oriented future – and cloud computing is one of the hottest trends. Cloud computing provides a single point of access for an enterprise’s computing needs, allowing them to cut costs and avoid unnecessary resources.

For all its benefits, cloud computing also presents infrastructure management challenges such as availability, security, policy compliance and support. To combat these challenges companies need to understand their IT infrastructure before – and after – moving to the cloud.

Maximizing the cloud requires a certain level of IT transparency and a thorough understanding of each infrastructure component and how it relates to others. Companies first need to discover and map their IT assets and identify crucial inter-system dependencies and relationships, ensuring all systems and applications are running smoothly and risk is mitigated. This type of data center search and analysis helps companies make the transition to the cloud fast and easy. And the ability to keep up with these types of changing IT infrastructures means users and cloud providers can detect irregularities, guarantee service levels and quickly identify risks.

The cost savings that cloud computing represents can be significant, and many organizations are set to embrace the model in 2009. But remember: IT shops are still quite traditional. They have infrastructure and enterprise management tools and the challenge is how to manage these services deployed on this nebulous cloud floating out there somewhere.

The problem with moving your services to the cloud is that you still own the machine images, so you still need to manage them – upgrading software, installing patches and so on. It also means you still need to monitor them. I was speaking with Mark Hall of Computerworld yesterday and we got onto the subject of “managing the cloud” … it occured to me that the traditional systems management tools, particularly monitoring tools, aren’t cut out for cloud computing. Most of these systems have agents which you need to deploy on your servers – that’s fine, you can still install them on your machine images in the cloud. However, many of these systems report status back to a central server or console, and now these things are out somewhere on the Internet. I can’t see your Security teams being too impressed about opening up those ports on firewalls – Imagine having to open UDP port 161 for SNMP traffic! An obvious fix would be some form of VPN, but thats an additional license cost per machine image.

Even a basic task like monitoring takes on new aspects when you are dealing with cloud computing. Not only are you dealing with traditional systems monitoring events but you have to watch for events from the cloud computing environment itself. For example, you want to be prepared when you get a message from a SaaS provider that its system will be down for maintenance or it’s moving to a new version. When they make those changes, what are they changing and which of your systems that integrate with that might be affected? And what do you need to do about it?