In an IT world increasingly dominated by vast amounts of data, it is becoming ever more important to accurately monitor, track and understand the statistics available. As companies migrate some of their services and applications onto the cloud whilst still retaining physical servers and databases, it is progressively more difficult to assimilate all the different sources and produce unified overview of the entire infrastructure.
A typical SaaS company uses such a wide variety of platforms that each department will have their own developers and system administrators using different monitoring solutions to gain the required insight on the data metrics. The wide scope of data sets being produced by incompatible monitoring tools results in a cumulative output that is difficult to decipher. Ultimately this results in a greater chance of misinterpretation and an increased likelihood of problems arising.
Providers need to maintain high levels of availability for applications and websites – outages and performance degradations lead to revenue loss, customer dissatisfaction and wasted productivity. If a problem occurs developers and system admins need to be able to rapidly detect, diagnose and resolve issues – that is impossible without effective data monitoring.
It’s not just large companies that need effective monitoring solutions. For small start-ups it is imperative to define monitoring systems in advance of the launch of a SaaS or PaaS. A successful service can jump from barely any users to tens of thousands in a matter of days. A famous example is that of Animoto – a video service that jumped from 5,000 users to 750,000 users thanks to a Facebook frenzy. The internal reality of a situation such as Animoto’s is that without effective monitoring the key personnel from other areas will be required to help developers and system admins deal with errors, alerts and bugs in order to keep the product live.
The only effective way to monitor data and provide a holistic overview of an entire company’s infrastructure is to use a solution that provides instantaneous system metrics, changes, alerts and events from across multiple sources. This allows developers and admins to use one portal and collaborate together in real-time, consequently reducing outages, improving performance, and saving time.
The challenge is to find a solution that can analyse raw metrics from a variety of sources. Datadog, a New York-based company founded in 2010, neatly solves this challenge because their industry-leading monitoring tool works with the most widely-used software, platforms, and languages that companies already utilise.
The service’s main strength is providing turn-key integration that enables users to take raw metrics from a range of outputs and turn them into meaningful performance data, interactive dashboards, and easy to manage alerts. A huge number of services are supported, from cloud platforms like AWS to automated configuration management platforms such as Puppet and Chef, and from SQL and NoSQL databases to web servers like Cassandra, Redis, and AWS Elasticache and DynamoDB.
Although Datadog was initially conceived as a monitoring platform for cloud servers, the solution is equally adept at overseeing on-site servers – it now monitors 20,000 cloud servers and 20,000 on-site servers daily. Since their launch the company has provided services for several high-profile clients, including HP cloud services, Intel, Salesforce.com, and Zendesk – meaning you can be sure of receiving a high quality service.
If you are developing a new app or platform it is essential to integrate a high-quality monitoring component to allow you to identify problems, optimise code, and perfect configurations. Similarly, if you have already developed existing software then using a monitoring tool that can seamlessly integrate with your all current IT infrastructure will reduce downtime, improve performance and increase revenue.
What do you think? Do you already use a monitoring tool? Can it integrate as flawlessly as Datadog or do you struggle to produce meaningful outputs?
By Daniel Price
