Promise of Cloud Computing
Cloud computing has revolutionized the way software is developed, tested and released. It has brought down overall cost and greatly reduced the effort and time to stand up the infrastructure. More and more enterprises are embracing cloud technology and deployment in cloud in growing at a much faster rate.
While some of the benefits of cloud computing is undeniable many organizations have suffered due to the lack of control and understanding of the underlying infrastructure. Things like am I adequately provisioning, do I need additional capacity, is my application problem due to infrastructure, can I see my application in context of my infrastructure, can I see bring all these monitoring to one place instead looking at 5 different monitoring tools and dashboards etc etc
Multitudes of Tools, Teams and Data Points
Let’s look at a scenario say TixChange, the online event company, recently launched a business critical and revenue critical application. Given the criticality cloud infrastructure was over-provisioned to ensure resource availability. ‘course this will cost them a “bit” more but at least they have peace of mind. As the launch and marketing campaign generated a huge buzz things started to get bit sticky. TixChange has some monitoring tools that provide them with some information around performance problem but is not able to point to anything definitive. So TixChange decides to look at some of the other monitoring provided by the cloud provider. ‘course that is a completely separate tool managed by a different cloud infrastructure monitoring team. The cloud infra team sees some issue but there is no way to tie the infrastructure with the application components that are deployed. While somebody is trying to make sense out of all these disparate tools, dashboards and data the customer is having a bad experience. All your time is spent on MTTI (Identification). The social media is abuzz with angry customers. An executive decision is taken by Matt to increase the capacity by order of magnitude, even if there was nothing to indicate definitively, with the hope that it would help “fix” the issue.
As you can see there are multitudes of issue with this approach:
Over-provisioning: to start with, as they have no good metrics to understand workload.
Unrelated Data: Even the data they have did not have a good correlation with the application behavior.
Data Proximity: The other challenge was the proximity of various pieces of information, managed by different teams, necessary to do effective triaging and multiple teams involved.
What TixChange needed was information that is current, accessible and correlated. A visualization of which is accurate representation of the physical layout and one team that had a complete 360-degree view of the everything.
Bringing a Sense of Sanity – Information at your fingertip
This is where CA’s new AWS field pack that is build on open source technology and integrate nicely with CA APM product helps companies bring both infrastructure and application key performance indicators together to provide a unified view that helps the monitoring team to make critical decision quickly and effectively. The field pack written in windows PowerShell with powerful AWS PowerShell API monitors all your AWS services like EC2, EC2 Tags, DDB, ELB etc. You can extend many monitoring aspects with simple configuration tweaks without any scripting skills. It not only publishes these metrics to CA’s EPAgent but also publishes key attributes into APM Team Center. This allows you to arrange your perspective that models your real life application layout allowing you to quickly determine if EC2 Instances on a particular availability zone is having a performance issue or its EC2 instances on all availability zone.
You can quickly use the right side attribute detail pane, which holds all key information in one place including things like location of these EC2, Maintenance windows, support contact etc allowing you to quickly triage the matter to the right team.
With metrics being published as EPA metrics you can now set alerts on EC2 state, any alarms that gets triggered by EC2 itself or any other kind of metrics. Since all these pieces of information are in ONE CA APM console your admin now has the ability make informed decision without looking over the shoulder.
And The Obvious…
The overall effect of this is reduced MTTR (repair), happy customers and business that doesn’t have to waste money on blind provisioning and guesses.
With the new AWS Field Pack and CA APM TixChange is now a happy camper seeing thousand-fold increase in their business without having to suffer like before