By John Dixon, Consulting Architect
With the acceptance of cloud computing gaining steam, more specific issues related to adoption are emerging. Beyond the big-show topics of self-service, security, and automation, cloud sprawl is one of the specific problems that organizations face when implementing cloud computing. In this post, I’ll take a deep dive into this topic, what it means, how it’s caused, and some options for dealing with it now and in the future.
Cloud Sprawl and VM Sprawl
First, what is cloud sprawl? Simply put, cloud sprawl is the proliferation of IT resources – that provide little or no value – in the cloud. For the purposes of this discussion, we’ll consider cloud to be IaaS, and the resources to be individual server VMs. VM sprawl is a similar concept that happens when a virtual environment goes unchecked. In that case, it was common for an administrator, or someone with access to vCenter, to spin up a VM for testing, perform some test or development activity, and then forget about it. The VM stayed running, consuming resources, until someone or something identified it, determined that it was no longer being used, and shut it down. It was a good thing that most midsize organizations limited vCenter or console access to perhaps 10 individuals. So, we solved VM sprawl by limiting access to vCenter, and by maybe installing some tools to identify little-used VMs.
So, what are the top causes of cloud sprawl? In IT operations terms, we have the following:
- Self-service is a central advantage of cloud computing, and essentially cloud means opening up a request system to many users
- Traditional IT service management (a.k.a. ITIL) is somewhat limited in dealing with cloud, specifically configuration management and change management processes
- There remains limited visibility into the costs of IT resources, though cloud improves this since resource consumption ends up as a dollar amount on a bill…somewhere
How is Cloud Sprawl Different?
One of the main ideas behind cloud computing – and a differentiator between plain old virtualization and centralization – is the notion of self-service. In the language of VMware, self-service IaaS might be interpreted as handing out vCenter admin access to everyone in the company. Well, in a sense, cloud computing is kind of like that – anyone who wants to provision IaaS can go out to AWS and do just that. What’s more? They can request all sorts of things, aside from individual VMs. Entire platform stacks can be provisioned with a few clicks of the mouse. In short, users can provision a lot more resources, spend a lot more money, and cause a lot of problems in the cloud.
We have seen one of our clients estimate their cloud usage at a certain amount, only to discover that actual usage was over 10 times their original estimate!
In addition, cloud sprawl can go in different directions than plain old VM sprawl. Since there are different cloud providers out there, the proliferation of processes and automation becomes something to watch out for. A process to deal with your internal private cloud may need to be tweaked to deal with AWS. And it may need to be tweaked again to deal with another cloud provider. In the end, you may end up with a different process to deal with each provider (including your own datacenter). That means more processes to audit and bring under compliance. The same goes for tools – tools that were good for your internal private cloud may be completely worthless for AWS. I’ve already seen some of my clients filling their toolboxes with point solutions that are specific to one cloud provider. So, bottom line is that cloud sprawl has the potential to drag on resources in the following ways:
- Orphaned VMs – a lot like traditional VM sprawl, resulting in increased spend that is completely avoidable
- Proliferation of processes – increased overhead for IT operations to stay compliant with various regulations
- Proliferation of tools – financial and maintenance overhead for IT operations
Download John’s ebook “The Evolution of Your Corporate IT Department” to learn more
How Can You Deal with Cloud Sprawl?
One way to deal with cloud sprawl is to apply the same treatment that worked for VM sprawl: limit access to the console, and install some tools to identify little-used VMs. At GreenPages, we don’t think that’s a very realistic option in this day and age. So, we’ve conceptualized two new approaches:
- Adopt request management and funnel all IaaS requests through a central portal. This means using the accepted request-approve-fulfill paradigm that is a familiar concept from IT service management.
- Sync and discover. Give users the freedom to obtain resources from the supplier of their choosing, whenever and wherever they want. IT operations then discovers what has been done, and runs their usual governance processes (e.g., chargeback, showback) on the transactions.
Both options have been built in to our Cloud Management and a Service (CMaaS) platform. I see the options less as an “either/or” decision, and more of a progression of maturity within an organization. Begin with Option 2 – Sync and Discover, and move toward Option 1 – Request Management.
As I’ve written before, and I’ll highlight here again, IT service management practices become even more important in cloud. Defining services, using proper configuration management, change management, and financial management is crucial to operating cloud computing in a modern IT environment. The important thing to do now is to automate configuration and change management to prevent impeding the speed and agility that comes with cloud computing. Just how do you automate configuration and change management? I’ll explore that in an upcoming post.
- Govern cloud without locking it down: see how AWS transactions can be automatically discovered by IT operations
- Influence user behavior: see how showback reports can influence user behavior and conserve resources, regardless of cloud provider
- Gain visibility into costs: see how IaaS costs can be estimated before provisioning an entire bill of materials
Register for our upcoming webinar being held on May 22nd @ 11:00 am EST. "The Rise of Unauthorized AWS Use. How to Address Risks Created by Shadow IT.