Imad Mouline is Chief Strategy Office with Everbridge in the US and has worked on ground-breaking technology throughout his career. In this article, Imad discusses how to measure the cloud so that it can be effectively managed.
As a former CTO for Gomez, I've lived and breathed application performance metrics on a daily basis. But more recently, I've been thinking about how to go beyond the technical metrics and measure the real information that matters to the business.
The concept of Business Services Management has been a buzzword in IT departments for years. The idea that technical metrics are not appropriate for business decisions is not a new one. You could even say it's like apple pie or motherhood - it's been around since time began. But before BSM was developed, acquiring the business implications of these metrics was a long process that required an enormous amount of analysis.
Directly applying business metrics to technology and infrastructure was a revolutionary step in IT management. By creating systems that could analyze technical information (server uptime, response time, traffic load, etc.) and convert them into information that mattered to business leaders (can customers still place and order? Is our internal portal functioning for all employees?), BSM drastically cut down the decision-making time needed for IT infrastructure projects. Whereas IT professionals previously scrambled to respond to every abnormal IT event, BSM now allowed them to prioritize their time and efforts to issues that had real and immediate business implications.
It came as no surprise that business leaders across the globe were hungry not only for the information BSM gave them, but for the cost efficiencies and strategic benefits the system brought.
So, what's the next step?
Today, businesses across the world are entering a new phase in IT infrastructure deployments: the cloud. Not only does cloud management provide real-time business metrics on which to base important IT decisions, it provides a hugely expanded level of analysis, flexibility, scalability and control. If BSM was the first step, then cloud management is its evolutionary successor.
Controlling costs and cloud chargebacks is, naturally, a key factor. Technologies such as CloudControl by CloudFloor and Cloud Cruiser are particularly effective at addressing cost management, albeit in different ways. Cloud Cruiser supports many cloud providers, including Windows Azure, and seems to specialize in cloud chargeback management. But CloudControl - the technology I lead the development for and I'm most familiar with - manages the resources, and hence costs, directly, by allowing IT executives and business leaders to tell their cloud deployments what capabilities are important and when and under what conditions they should be changed. For example, should a traffic spike occur, you can program the infrastructure to increase the scope of its deployment; in this way it can handle the increase safely without any performance degradation. Conversely, in periods of time that are particularly slow, the infrastructure can automatically scale itself back, preventing any unnecessary costs.
Some cloud management systems also have advanced DNS capabilities built-in, allowing them to use geographic information to alter deployment. If servers are more expensive when hosted in a particular region, you can set up the infrastructure to use those options only when absolutely necessary. Alternatively, if there is a particular time of day that is low traffic, you can allocate cheaper (potentially less capable) servers for those periods and so save costs. But the main application of advanced DNS capabilities is in deploying end-user traffic and requests to the most efficient or appropriate cloud resources - in real-time. So if a particular cloud region or availability zone starts showing signs of poor stability, the cloud control mechanism uses the DNS capability to direct incoming user traffic away from the seemingly unstable cloud region, and toward a different set of cloud resources. When combined with direct control over the cloud resources, the control mechanism can also quiesce or ramp down the unstable resources (to save costs), and ramp up the alternate set of resources. It is this ability to direct end-user or application traffic, while managing the resources to which it is directed, that makes a cloud management solution like CloudControl so powerful.
Through the cloud, you can now see how much a particular infrastructure set-up impacts on cost per conversion, whether page load time affects revenues, what circumstances cause people to abandon incomplete transactions, and so on. In this way you can not only see what customers or users are able to do, but whether there is a more efficient way of doing it.
It can get even more specific than that. Not all businesses value the same metrics equally; because it's so flexible, cloud management software can be tailored to provide companies with exactly what they are looking for. For example, a large auction house might always want to know the number of bids or the ratio of bids to revenue; an image hosting site might want to know the total number of uploads or total number of views; a music streaming website might want to know how many minutes of music are uploaded every second, and so on. In essence, this allows you to program your infrastructure to meet the specific conditions of your SLA in the most efficient manner and at the lowest possible cost.
Cloud management technology helps prioritize IT projects by correlating available IT information with relevant - and corresponding - business goals. Additionally, it allows the infrastructure itself to scale to those business goals and needs. Who knows - in years to come we may look back at these early days of cloud computing, and wonder how we managed without easily accessible business services metrics, a bit like life in the 80s before mobile phones!