Optimize Two Outcomes: Happy CFO & Happy User
With all the turmoil in the data center with changes in both the server and desktop virtualization landscapes, it is not surprising that IT is focused on reducing cost. Additionally, VDI has been associated with poor user experiences, stemming from poor performance because of latency and challenges with running a complex system. These are the 2 outcomes we are focused on optimizing: reducing cost and delivering a better user experience.
Historically VDI has been an expensive data center workload requiring lots of infrastructure and specialized teams for support. The common perception has been that the cloud is more expensive to deliver VDI. However, with proper optimizations, the cloud can be significantly cheaper. Desktops are an embarrassingly cloud-native workload – You can closely tie usage to capacity (and cost) for each user. So start/resume the resource when the user needs it and stop/pause the resource when the user does not, thereby optimizing cost.
Similarly with the availability of almost infinite storage in the cloud, we can collect a lot of data in real-time to understand and diagnose end user experience, thereby delivering the best possible end user experience.
Use AI to Optimize Outcomes
AI is a fantastic tool to help us in both cost optimizations and delivering a better user experience.
For cost optimizations, AI helps us:
- Forecast usage of when users are logging in and accessing their virtual resources
- Just-in-time provisioning to ensure that the virtual resource is ready just when the user needs it, and
- Just-enough provisioning to ensure that we don’t over-allocate resources
To deliver the best end user experience, AI helps us:
- Analyze user sentiment by correlating sentiment to metrics
- Accelerate root cause analysis by looking at previous alarms (along with metrics, logs and traces) and their root causes to analyze future alarms
It All Starts with Data, Data, Data
It all starts with data – metrics, logs and traces. From all over:
- From any device
- From the networking gateway
- From the cloud or data center
The amount of data is huge – we collect multiple data points from every data point every 30 seconds. We also collect logs, traces, etc. from all these and all our DaaS system components.
User-First Data
The data needs to be user-focused. Data from some systems that are shared by multiple users needs to be adjusted to make it user-centric so we can look at the data in the context of single user, single pool of users, etc.
Global Time Sorted
Since we are collecting data from many distributed systems across the globe, it is important for the data to be sorted by time, so that we can do root cause analysis more effectively.
Real-time, Analytics, and AI
Once the data is available, there are multiple ways to utilize the data:
- Real-time Alarms
- Analytics
- AI
