Is the Cloud a fad?
A fashionable term for several years, the Cloud has become an essential component of any IT infrastructure. The promises are numerous: increased performance, high availability, accessibility, flexibility, cost reduction, scalability, security...
Driven by giants such as Microsoft, Google and Amazon, cloud platforms are maturing and allowing production use.
Indeed, companies are increasingly turning to hybrids (some applications in the cloud and some in-house) and multi-cloud (some of their applications are hosted in a public cloud while others remain on internally managed servers and are delivered via the corporate network).
Platforms such as Azure or AWS provide a wide range of resources in IaaS or PaaS mode that it is essential to monitor.
PaaS/IaaS, what are the differences?
Let's take a concrete example.
A company wants to create an application to collect data sent by connected objects (IoT), process them, store them and display them on a web application.
In the environments Azure and AWS2 paradigms coexist: the IaaS and the PaaS.
The IaaSfor Infrastructure as a ServiceThis service allows you to rent virtual servers in the same way as with any other hosting company.
The machines are accessible via a console or a remote desktop and their administration (update, configuration, installation of services) is done like a classic virtual machine.
To create its IoT application, the company can decide to rent 2 servers in IaaS mode.
In addition to managing the OS and the machine's libraries, it will also be necessary to keep each service up to date individually.
The services will be set up from the servers via their respective configuration files.
The PaaSfor Platform as a Servicepresents a new way to design and manage an application architecture.
This type of cloud computing allows companies to focus solely on the development and configuration of an application without the need to deploy and maintain the infrastructure.
The resources made available by these environments are not virtual machines but micro-services.
Let's go back to our IoT application.
In PaaS mode, the company does not use virtual servers but PaaS resources.
The different functionalities of the application are no longer shared on 2 servers but are split into several micro-services, each rendered by a PaaS resource.
Each service can be configured via the web interface of the cloud platform. The underlying infrastructure is totally abstracted, we no longer deal with physical machines, OS, libraries, updates but only with the configuration of each resource.
This example shows an architecture on Azure, but AWS allows the same type of application.
What about supervision?
Let's continue with our IoT application and add local network monitoring.
An efficient supervision will be necessary to avoid any unavailability and to keep a maximum of visibility on the whole information system.
It will be based on two needs:
- Monitoring of classic servers in the local network with SNMP and/or WMI protocols.
- Monitoring of PaaS resources through APIs.
ServiceNav allows you to monitor both environments and aggregate the results on a single web interface.
Azure and AWS PaaS monitoring
The use of PaaS resources allows you to free yourself from the system part and focus on the configuration and development of the application. However, whether for AWS or Azure, it is still essential to collect metrics to measure usage and application performance.
Azure and AWS have their own monitoring solutions: Azure Monitor for Microsoft and Amazon Cloudwatch for AWS.
Directly integrated into the web administration interface, they have the advantage of being very complete but may not be suitable for all users.
They will keep developers happy with a multitude of technical dashboards and general application performance metrics, but may be complicated for support teams to use, especially in a multi-client, multi-cloud context.
Our plugins (MS-Azure-PaaS-Metrics_v2 and AWS-Paas-cloudwatch) rely on the APIs provided by Amazon CloudWatch and Azure Monitor to feed data back into ServiceNav.
Despite the PaaS/cloud dimension, the information will be treated as "classic" monitoring data and will benefit from all the advantages offered by ServiceNav: comparison with thresholds, maintenance, acknowledgements, ITSM integration, time slots, alerts, data historization, average and peak hour calculations, display in Dataviz and soon forecasting.
By monitoring precise metrics integrated with weather servicesServiceNav makes it easy to track and resolve an incident by directly pointing to the impact of an alert on the entire application.
The display on Dataviz will also give a complete view of the application: