Forecast: Cloudy, with growing demand for insight

With the help of sophisticated aviation systems, today’s pilots are able to fly through dense cloud cover. Instruments provide pilots with visibility that their eyes cannot. But imagine what it must have been like for pilots before the advent of modern navigation systems. On a clear day, you could see forever. On a cloudy one, you might find yourself flying blind.

Ask any IT department running workloads in a public cloud environment, and they can probably tell you a thing or two about what it means to “fly blind.” While there is tremendous efficiency and elasticity to be gained by migrating workloads to a hybrid or public cloud environment, there are also major drawbacks. IT teams lack of visibility into cloud-based applications in use, resources consumed, and end-user experience, resulting in significant challenges in troubleshooting and capacity planning. This article will look at three areas where added visibility can improve public cloud deployments, and provide a checklist of what to look for in a cloud monitoring solution.

1. Making the Cloud Cost-Efficient Requires Visibility

One of the major benefits of migrating applications to a public cloud environment is the ability to automatically provision new compute resources as demand increases. However, when errors cause an otherwise unnecessary auto-scaling event, it can be costly.

Conga, a leading Salesforce application partner, knows this pain first-hand. In order to keep pace with its rapid growth, Conga migrated its applications to AWS. While AWS afforded Conga much greater scalability, it also left the Conga team with limited insight into application performance. The company found that costly auto-scaling events were often the result of unexpected application behavior, rather than increased customer demand. This left Conga out for the cost of the additional (and ultimately unnecessary) compute resources.

2. Protecting End-User Experience

When Conga migrated their Salesforce applications to AWS, they gained elasticity and scalability to support their more than 100,000 users in 35 countries; however, they also lost visibility into the behavior of their applications, especially with issues involving Active Directory, DNS, and load balancing. For Conga, this resulted in serious challenges. Not only was it difficult to determine the source of a performance problem—whether it was stemming from the Conga application, from AWS, or from Salesforce—they also lost visibility into real-time end-user experience. Rather than being able to troubleshoot a problem before it caused major service disruptions or slow-downs, Conga regularly found itself engaged in a fire drill, trying to troubleshoot the root cause of the latency or error as end-users languished.

3. Keeping Cloud Providers Honest

Any cloud-based service, whether it’s SaaS or IaaS, enters into Service Level Agreements (SLAs) with its customers specifying things like availability and Quality of Service (QoS). Without the ability to independently monitor the performance and availability of cloud workloads, however, IT is often dependent on the cloud provider itself to tell them whether SLAs commitments are being met.

Having the right cloud monitoring and visibility platform returns control to the IT department, delivering independent insight into availability and QoS that keeps cloud providers honest.

What to Look for in a Cloud-Monitoring Platform

The demand for operational visibility in public and hybrid cloud environments has lead many to try to have force-fit traditional monitoring tools into virtual packages, with little or no attention paid to the unique attributes of cloud environments and services, or the need for visibility across cloud and on-premises deployments.

Conga’s search for an application monitoring solution for AWS was arduous given the complexity of their application, which relied on “merge” requests comprised of multiple asynchronous flows between different services. This was not something that was easily tracked with log monitoring, agents, or resource utilization. Eventually, Conga found a solution that could provide the detailed visibility in AWS that they required, leveraging network or “wire” data that could provide insight despite the scale, complexity, and dynamism of their cloud deployment. Now, they can see what is going on with their application that is causing spikes in AWS usage as well as identify specific users that are experiencing slowness or application errors.

Extrapolating from their experience, here are some key attributes to look for when considering a cloud-monitoring platform:

ü Real-Time Visibility. Learning about a costly autoscaling event when the bill arrives isn’t particularly helpful. The damage is already done. The best cloud monitoring solutions provide not only real-time insight that allows IT teams to spot unusual behavior early, but the details needed to troubleshoot and resolve the problem quickly before new compute resources are spun up.

ü Datacenter Scale. Even as datacenter speeds continue to increase, with 10GbE shipments now leading the market and 40GbE in the early stages of gaining broad adoption, the ability to monitor at these scales in virtualized and cloud environments has lagged far behind. Many virtual monitoring offerings lag far behind their physical counterparts. When considering a monitoring solution, it’s critical to understand which virtual offerings will be able to monitor at scale in the most cost-effective manner.

ü Holistic Infrastructure Monitoring. While enterprises have embraced both public and hybrid cloud in recent years, many workloads still remain on-premises, and most IT infrastructures leverage some combination of private cloud, public cloud, and traditional datacenters. A solution that provides visibility only into the cloud or virtualized environments, and a separate tool that provides visibility for on-premises deployments, can leave IT with significant blind spots. Look for platforms that can monitor across all deployments.

ü A Multi-Source View. As enterprise IT has learned over and over again, the promise of a “single pane of glass” through which to monitor the IT environment most often results in a “single glass of pain.” Especially in complex, dynamic, and hybrid environments, a single data set isn’t going to provide insight into every single nook and cranny. When considering a cloud monitoring solution, it’s important to consider not only what data is being used in the performance analysis, but whether the product allows that data to be combined with other sources for more comprehensive insight.

As hybrid and public cloud deployments continue to increase, enterprises will need to find ways to regain the visibility they enjoyed with on-premises deployments. In the meantime, many IT teams will be flying blind. The key to accelerating cloud adoption is to provide enterprise IT with the technologies it needs to navigate these cloud environments.