VMware – VMworld 2019 – HBI3516BUS – Scaling Virtual Infrastructure for the Enterprise: Truths, Beliefs and the Real World

Disclaimer: I recently attended VMworld 2019 – US.  My flights and accommodation were paid for by Digital Sense, and VMware provided me with a free pass to the conference and various bits of swag. There is no requirement for me to blog about any of the content presented and I am not compensated by VMware for my time at the event.  Some materials presented were discussed under NDA and don’t form part of my blog posts, but could influence future discussions.

These are my rough notes from “HBI3516BUS – Scaling Virtual Infrastructure for the Enterprise: Truths, Beliefs and the Real World” was a sponsored panel session hosted by George Crump (of Storage Switzerland fame) and sponsored by Tintri by DDN. The panellists were:

JP: Hyper-V is not really for the enterprise. Configuration, and automation were a challenge. Tintri made it easier to deal with the hypervisor.

JD: You put a bunch of disks and connect it up to what you want to. It’s really simple to setup. “Why would you want to go complex if you didn’t have to?”

MB: When we had block storage, we were beholden to the storage team. We’ve never had problems with their [Tintri’s] smallest hybrid arrays.

AA: Back in the ESX 2.5 days – single LUN per VM. We would buy our arrays half-populated – ready to grow. We’re now running 33 – 34 devices. Tintri was great with QoS for VMs. It became a great troubleshooting tool for VMware.

GC: Reporting and analytics with Tintri has always been great.

MB: We use Tintri analytics to create reports for global infrastructure. Tintri will give you per-VM allocation by default. Performance like a Tivo – you can go back and look at analytics at a very granular level.

GC: How did the addition of new arrays go with Global Center?

MB: We manage our purchases based on capacity or projects. 80 – 85% we consider additional capacity. Global Center has a Pools function. It does a storage vMotion “like” feature to move data between arrays. There’s no impact.

JP: We used a UCS chassis, Tintri arrays, and Hyper-V hypervisor. We used a pod architecture. We knew how many users we wanted to host per pod. We have 44000 users globally. VDI is the only thing the bank uses.

AA: We’re more of a compute / core based environment, rather than users.  One of the biggest failings of Tintri is that it just works. When you’re not causing problems – people aren’t paying attention to it.

MB: HCI in general has a problem with very large VMs.

AA: We use a lot of scripting, particularly on the Red Hat (RHV) side of things. Tintri is fixing a lot of those at a different level.

GC: What would you change?

JP: I would run VMware.

MB: The one thing that can go wrong is the network. It was never a standardised network deployment. We had different network people in different regions doing different things.

JP: DR in the cloud. How do you do bank infrastructure in the cloud? Can we DR into the cloud? Tested Tintri replicating into Azure.

AA: We’re taking on different people. Moving “up” the stack.

Consistency in environments. It’s still a hard thing to do.

Wishlist?

  • Containers
  • A Virtual Appliance

 

Thoughts

Some folks get upset about these sponsored sessions at VMworld. I’ve heard it said before that they’re nothing more than glorified advertising for the company that sponsors the session. I’m not sure that it’s really any different to a vendor holding a four day conference devoted to themselves, but some people like to get ornery about stuff like that. One of my favourite things about working with technology is hearing from people out in the field about how they use that technology to do their jobs better / faster / more efficiently.

Sure, this session was a bit of a Tintri fan panel, but I think the praise is warranted. I’ve written enthusiastically in the past about how I thought Tintri has really done some cool stuff in terms of storage for virtualisation. I was sad when things went south for them as a company, but I have hopes that they’ll recover and continue to innovate under the control of DDN.

When everything I’ve been hearing from the keynote speakers at this conference revolved around cloud-native tools and digital transformation, it was interesting to come across a session where the main challenges still involved getting consistent, reliable and resilient performance from block storage to serve virtual desktop workloads to the enterprise. That’s not to say that we shouldn’t be looking at what’s happening with Kubernetes, etc, but I think there’s still room to understand what’s making these bigger organisations tick in terms of successful storage infrastructure deployments.

Useful session. 4 stars.