Day 4 of the VMworld conference was filled with even more exciting sessions and announcements. As we walked through the exhibit hall, we couldn’t help but feel at home with all the familiar faces and solutions on display. While I didn’t have as many sessions scheduled for today, I still managed to learn a lot and discover some innovative solutions.

One of the highlights of my day was attending the “Clustering Deep Dive” session by Duncan Epping and Frank Delenia. With adoption rates of over 90%, clustering is clearly a popular feature within the VMware ecosystem. As administrators, we have become accustomed to restarting hosts within a VM after a hypervisor crash, but it’s important to remember that this process should be tested and not taken lightly. Duncan emphasized the importance of thoroughly testing all configurations before deploying them in production, as some parameters can have disastrous consequences on the resilience of your infrastructure.

The session also delved into DeepDive’s analysis of logs within an ESXi cluster during incidents such as hardware failures. Proactive HA, which allows for the analysis of degradation in equipment health to anticipate failures and trigger preemptive migrations, was also discussed. While this technology has been around since 2006, it is still not widely deployed. Frank provided a detailed explanation of the various metrics used to assess the health of VMs, their neighbors, and their future state when considering a migration.

DRS (Distributed Resource Scheduler) was another topic that was covered in depth. With its numerous configuration options, DRS can be optimized for placement within a cluster. The session highlighted the importance of CPU %ready, memory active vs. consumed, distribution of VMs, and network-aware DRS. Additionally, the settings VM distribution, which attempts to prioritize “VM Happiness,” has three primary use cases:

Finally, the Predictive DRS (associated with vROPS) was discussed, which allows for the prediction of behavior over the next 60 minutes and considers the current happiness state of a VM when deciding whether to migrate it.

In summary, day 4 of the VMworld conference was filled with valuable insights and innovative solutions within the VMware ecosystem. From clustering to DRS and predictive analytics, there is always something new to learn for administrators looking to optimize their virtualized infrastructure.