(Originally written for SRI Infotech. Republished with permission.)
The cloud is becoming an increasingly popular way for companies across all industries to reduce costs related to their IT infrastructure. By embracing the cloud, companies are able to reduce technology infrastructure costs, innovate more rapidly and engage in more responsive client relationships.
However, operating in the cloud requires a different approach to monitoring and optimization.
Why Monitoring Must Change
In the past, applications often ran on specific hardware, physically separated from other dissimilar IT resources. Different technology, languages and platforms were common, but generally of little consequence. Siloes were frequent, but didn’t impact monitoring, load or performance. Each team developed monitoring techniques and skills expertise uniquely beneficial to their specific needs.
In today’s cloud environment, different applications using different platforms and different technologies are often in the same stack. This differentiated environment requires coordination between team owners that was previously unnecessary.
In a traditional data center, it was common and acceptable for development teams to generally ignore infrastructure. Dedicated server teams took responsibility for the physical environment, while application teams monitored app and end user performance. However, in a cloud scenario, development teams have end-to-end responsibility, which includes the backend infrastructure.
Why? Application changes – which happen more quickly in the cloud – may have unanticipated repercussions on application, environment or end user performance across the cloud. Degraded performance is no longer limited to one team. It can affect the entire organization.
As a result, there is now a need to facilitate communications between employees responsible for host, network, application and end-user monitoring. From service level data, such as errors or response times, to real-user monitoring, including end-user response times and errors to behavioral data and log monitoring need to be freely available amongst unrelated entities. Without improved communications strategies, transparency and visibility will be lacking, and problems increasingly likely.
Potential solution – Big Data
In order to facilitate the collection and sharing of performance metrics at each particular layer (infrastructure, application, user), organizations must invest in both automation and technical expertise. It is unlikely that a single tool will suffice to bring necessary data points together. However, by feeding data streams from different solutions into a common big data back end, greater transparency is available for processing and analysis.
By implementing a big data monitoring protocol, insight into the health of the cloud environment becomes more accurate, more available through decreased collection cycles and more meaningful to a greater number of users in a shorter amount of time.
Moreover, because big data is uniquely qualified to consolidate and make sense of data points from numerous sources in various formats, it enables the detection, and even the prediction of performance issues.
Combining big data with cloud monitoring mitigates operational risk, reduces costs and helps maximize the potential of the cloud, ensuring return on investment and increased operational efficiency.