On July the 9th I attended the second London VMUG event based from OVH cloud’s offices. Here the low down on what occurred:
The first session was the welcome speech from VMUG leader Chris Bradshaw.
Chris ran through 10 feature updates for VCF 9.1 update:
- VCF Ops Logs
- Infrastructure policy
- K8 Cost Visibility
- IAM Enhancements
- Containers as a service
- Tag to objects
- VM Service Enhancements
- Memory Tiering Enhancements
- Realtime Metrics
- Appstack formation
VMUG Leader Gareth Edwards then ran through some further new features of note in 9.1
- vCenter quick patch
- ESX Live patching
- Desired State
- 25% Increase in scalability
- DRS optimised maintenance
- Improved vMotion resource utilisation
- Enhanced Memory Tiering
- Encrypted vMotion offload
- Optimised scale and performance for latest CPUs
- Unified identity
- DSM (Data Services Manager) enhancements
- Private AI enhancements (New HW support)
- GPU Observability
The next session was held by Vedat Kurshumliju (TAM, Broadcom) on driving greater efficiency with VMware VCF. In this session Vedat covered how he uses VCF Operations Aria Operations to look at capacity reclamation before he advises his customers to look at new hardware. He does this by using the following criteria:
- Aged Snapshots
- Orphaned VMs
- Powered off VMs
- VM Configurations
- Idle VMs
- Rebalancing workloads across clusters
- Right sizing VMs (oversized and undersized VMS)
He ran through how he advises his customers to methodically step resources down (or up) and test until the VM sits just right. He included examples where he has used this to help customers identify a bottleneck VM that Aria Ops suggested required more memory. This reduced the number of support tickets for this application causing slow performance. He also advised how he used automation to help his customers with the resourcing, for example he created a policy that would email users with a snapshot over a certain amount of days to advise that it is due to be deleted. He also advised of using a system with a team of developers that allowed them to see where they should best place any test VMs based on a traffic light system.
The next session was from Atif Qadeer (Broadcom) called Making every layer react: rethinking security in VMware Cloud Foundation. Atif explained that defence is like an onion (makes you cry with each large release). He explained that there were forgivable and unforgivable vulnerabilities and that the average dwell time of ransomware is around 204 days. He advised that most common applications tend to be grouped into similar groups and that once a threat takes hold it can remove elements of detection. Atif went through a DNA of ESXi networking including the increasing amount of traffic slots.
He also went through how the vDefend distributed firewall provides a highly scalable control plane with purpose built recovery, advanced security and hardened infrastructure with AI powered threat analytics. There was a live demo (that was recorded) on viewing an attack, lifecycle management, micro segmentation and reporting.
After a break, we moved onto a community session on using packer with Github with James Kilby (Broadcom). James covered how traditionally built VMs are slow to build, inconsistent, offer no audit trail and can be hard to reproduce and how he solves this problem using automation with code. He explained about Github actions and packer from Hashicorp and how he has a self hosted runner that allows the code to be run securely on his home environment. He ran through a live demo of using the GitHub actions to download an ISO file, add it to a content library and use it to build a VM.
He explained how if packer fails it will clean everything up and keeps the packer log to troubleshoot what went wrong in the workflow. He advised how this is a handy method for rotating templates in VDI environments and how to setup runner and add secrets to secure the connection between your environment and GitHub.
After a spot of lunch, we moved onto a session by Paul Toal (Broadcom) called security for AI driven enterprise. Paul advised how apps are evolving from human coding through to multi agent and how current LLM tiers have no security.
He explained how VCF Private AI is the missing layer with policy and governance. There was an example where he ran through a day in the life of an AI agent and how it has no guardrails to govern what it is doing. Paul went on to advise how we can use PEP gateways to enforce decisions and provide security, observability and identity to these services.
Then for the final session of the day we had Advanced Network Services for VCF with Dan Watson. Dan ran through ANS (application networking security) and how it helps to deliver, protect and observe.
He then ran through the 5 principles for using AVI load balancing for private clouds and some of AVI’s enterprise application delivery features.
After a final speech with thank you’s and dates for the diary, we all headed to the local Brewdog pub for food and drink and a few games of shuffleboard courtesy of OVHCloud who very kindly sponsored the event.
It was a really good day full of learning, networking and catching up with people.