Business-driven processes for measuring and guaranteeing service levels in the cloud will help CIOs successfully transition to this new environment. Monitoring public cloud applications is different from monitoring premises-based systems and brings new challenges that need to be addressed by an organisation.
One of the main reasons for monitoring applications in the cloud is to manage and maintain optimum levels of performance and availability. From an infrastructure perspective, the key performance indicator is device and resource utilisation.
An IP infrastructure needs to be able to support the prompt delivery of business applications from server to end user device, so any over utilisation of resources along that application path will impact the delivery and therefore the quality of those applications; hence it is vital to monitor these resources continuously, as well as keep a benchmark of historic utilisation.
Whilst, over utilisation is far more likely to impact service than underutilisation, it is essential to monitor both because underutilisation could provide you with an opportunity to right size your infrastructure and downgrade, or simplify, an associated service to reduce cost.
Due to the nature of SaaS (software-as-a-service) it can be difficult to effectively monitor cloud applications operating across the open public network; this is because as one of many clients sharing your SaaS provider’s resources, you will not be able to get visibility of the application path from server to end-user.
What can be done, however, is transaction or quality of experience testing. This involves monitoring the availability of the application at a defined interval and frequency, measuring the speed and responsiveness of the application by running transactions, and creating performance benchmarks; these can then be used to create alerts to highlight abnormal performance.
Once past the service demarcation of the SaaS multi-tenanted infrastructure, the rest of the application path, from the SaaS data centre to the end user device, can be monitored both accurately and efficiently.
When it comes to managing private clouds, the good news is that you should be able to manage and monitor the entire application path and, to make the process easier, there are now a number of off-the-shelf tools such as Solarwinds Orion, Ipswitch What’s Up Gold and CA eHealth, for example, that will help you carry out the task.
The key challenge, for anyone without specialist knowledge of infrastructure monitoring, is in understanding, out of all the performance data collected, what is really pertinent and a proven indicator to the quality of your application’s performance. What can help is to use a broad-based tool to give you an overview of your total IP infrastructure and then have a secondary level of device or application specific tools for deep dive analysis.
Most monitoring tools will also cover virtual systems, so virtualisation in both private and public clouds does not pose any new challenges although it is important to stress that the administrators of private clouds, with dedicated infrastructures, will have greater visibility than those working with public or shared virtual solutions. In shared or hybrid clouds, the level of visibility will depend on your supplier’s service demarcation and what they are willing to show you.
Unfortunately, it’s not all plain sailing from here. Load testing, which provides valuable benchmarks for application monitoring, is going to be a challenge for anyone subscribing to a public cloud offering. To undertake this kind of testing you need knowledge of the cloud system’s topology which you are unlikely to get from your SaaS provider, purely for security and commercial reasons. All you can really do in this situation is to monitor user login and transactional speeds.
As cloud applications move towards richer HTML5 clients, monitoring application performance will become more difficult because current monitoring tools are severely limited in this area. There are currently a number of tools available that have been specifically designed to monitor user experience based on web applications.
Solarwinds Orion is one such application; it achieves this via its ‘Synthetic End User Monitor‘, a feature that enables the recording of frequently used actions within a web site or page and allows issues to be reported against these actions.
One of the biggest, and most time consuming, challenges that organisations will face with the cloud is issue resolution. Pinpointing the cause of application performance issues and providing evidence of root cause, and therefore ownership, cannot be achieved purely by maintaining a tight grip on SLA dependencies between physical, virtual and cloud environments.
You need to have a complete view of the application path or as close as you can get. You then need to monitor the SLA of each device, each network segment and, the overall application path. Once this is done root cause analysis is completely viable.
Cloud applications offer a great opportunity for organisations to improve the flexibility and scalability of the IT services they deliver to end users. Application monitoring is not only possible in the cloud, it is vital, and should be a constituent part of every IT strategy that embraces cloud computing.