Seems like with cloud computing and the SaaS/PaaS models everyone is starting to really ask questions around availability and Service Level Agreements. And why shouldn’t they be? Moving your critical application functionality outside your firewall and onto the servers of a third party is a big step that makes you vulnerable to the downtime of systems you have little to no control over. Both Amazon and Twitter have had high profile downtime events recently which have caused users to really dig deep into the SLAs of these services.
The concept of monitoring performance of an application can become quite complex when third party services are used to provide key functionality. On the one hand you need to be able to drill-down on the various components that make up the application (previously it would have been sufficient to monitor the performance of the web based application from just an external users perspective, but now your servers are the users that are using third party components so that monitoring needs to be done from behind your firewall). In addition to the added complexity of end user perspective is the sharing of the data that is being collected. If you’re a producer of services then you want to be able to share your uptime and performance SLA statistics (creating an open and trusted relationship with your users) and if you’re a consumer of third party services then you will want to be sharing the appropriate performance data with various departments in your organization (as well as be able to present that data to your third party providers if SLAs are ever in question).
If you have a moment, check out the Webmetrics Eco-System monitoring platform for a solution to managing the SLAs of third party components and mashups. I would be interested to hear any feedback you have. I promise I will not focus this blog too much on Webmetrics based products and will try to keep it focused on web performance monitoring concepts and best practices. In this case I feel that keeping track of the availability and performance of third party components and services is a critical factor in managing any web application that utilizes third party services.
The performance for this blog:
- Average load time is 1.73 seconds.
- Availability (uptime) is 100%.
