Tuesday, July 18, 2017

.next 2017 in DC

Just a few observations on the Thursday keynote and a few sessions at the .next 2017 conference in Washington, D.C.

Location, Location, Location

The conference organizers picked a great spot for the event. The Nutanix .next conference is a young tech conference, and doesn't yet pull the massive crowds that established events can pull. But that's not a criticism, it's a compliment: VMworld and CiscoLive have grown into such massive events that it's easy to get lost in the swag-seeking throngs and never establish a connection with anyone. And yes, I know that many people will share their opinion that conferences are just opportunities, and taking advantage of these opportunities is ultimately up to the attendee. True, mostly. But I've always preferred a small group of similarly interested parties over 25,000 people bouncing from vendor party to vendor party.

The Gaylord National easily accommodated the group, albeit it with more than a few trips up and down the escalators. And as a resident of the greater Baltimore area, I found getting to and from the event to be surprisingly reasonable. One of the reasons I don't attend more of these events is less about the cost of attending and more about the time commitment. But I don't feel guilty about spending a long day or two learning and networking.

Finally, D.C. gets very few events like this that I'm interested in attending. So kudos to Nutanix for serving the tech community of the DMV.

VM... who?

It's no secret that the relationship between NTNX and VMW is wrought with years of competition and, in some cases, exceptionally crude name calling (tweeps, you know what I'm talking about). But since the Nutanix IPO, they're toned down their messaging, which is a very welcome change. I admire the tech of both companies, and have good friends who are committed to each. The vitriol was approaching "us vs. them" territory. But at .next 2017, VMware was only mentioned as a supported platform and partner for VDI solutions. I heard no back-handed compliments or snarky remarks. Instead, the focus was on the features of ACH, Xi, and Calm. We all know that certain individuals at Nutanix love to go competitive whenever they have the chance. But it's my opinion that customers and tech enthusiasts don't care for the sniping, and would prefer a conference that focuses on the strengths of a product, not the comparative value of solutions.

Cost

I don't attend many conferences these days due to my schedule, and it's rare that I can take an entire week off of work to fly out of town and rack up $5,000 or so in total attendance costs. But the .next organizers offered day passes, which turned out to be perfect for me. I was able to review the agenda before determining which day would be most relevant to my work, and buy a pass for that day alone. I caught some great sessions (including Chris Wahl's session on vester and automation in general), bumped into a few old friends (Saddler!), and only ducked out twice for work-related conference calls. The is a great option when the event is within a reasonable drive from home.

Friday, July 14, 2017

Updates from Ravello

Ravello Scooter, circa 2014
In 2014, Ravello Systems set up their booth at the VMworld expo hall and offered a pretty startling proposition: run your vSphere VMs in the cloud, and do it natively. It's that last bit that had everyone buzzing. We were all familiar with various methods for converting a VM from one virtualization solution to another, but we hadn't seen a working demonstration of forklifting a VM and dumping it in a cloud.

Also, they had scooters.

I stopped by the booth that year and had a chance to speak with some really smart and enthusiastic people about nested virtualization. And while it's been a few years, I'm confident that I spoke with Shruti Bhat, who presented Ravello's solutions to the tech bloggers at Virtualization Field Day in 2015. She set me up with some complimentary access to the solution (this was back when I was a #vExpert, after all), and I walked away from the booth pretty excited about where this tech would lead us.

Lots of home labs popped up in Ravello shortly thereafter. People were really excited about this stuff.

Skip ahead to 2016: Oracle buys Ravello, and we assume the worst: Ravello is the walking dead, a promising solution that's about to be assimilated into... something. Things get quiet, time passes. We pack away the scooters, neatly fold our "Clouds are for the birds... my VMs are nested" t-shirts, and move along.

The Rebirth of Slick

But now, Ravello is back in a big way. And not just with an exciting roadmap of features, but also with a clear direction and place in Oracle's cloud strategy. Here's a quick overview for you.

On-Prem to Cloud Migration

It's only slightly more complicated than this image suggests.
There's no point in offering a cloud solution if you don't give customers a really easy-to-use on-boarding process. Oracle Public Cloud is now more accessible thanks to Ravello, which can run your VMware VMs on AWS, Google, and now Oracle's Public Cloud. Smart move to bring a smart nested virtualization solution in-house to make cloud adoption that much easier.

The HVX Hypervisor

We don't spend a lot of time talking about Type 2 hypervisors these days, but HVX might change that. Initially, Ravello was a consumer of AWS and Google cloud resources, so they were required to build their solution within the elastic instances in the cloud. But Oracle's cloud removes that constraint, and we can expect to see a Type 1 version of HVX in a future release. If we've learned anything about virtualization, it's that the closer your hypervisor is to compute, memory, and storage, the faster your VMs will run. Expect major performance increases when HVX attains Type 1 status in Oracle's Bare Metal Cloud.

Blueprints

Personally, I cringe when I see the word blueprints used in cloud workspaces, because it makes me think of people who use the word "architect" as a verb. And it attempts to elevate routine tasks into something grand. Nevertheless, the word persists.

In Ravello's defense, they have a refreshingly simple interface to use when you are connecting your nested VMs using the magic of overlays. It's got a slick UX, too. Just connect your VMs to the network topology that you create in the cloud, and you're good. Yes, I'm oversimplifying. But not by much.



SDN, With Actual Purpose

The networking magic is what makes Ravello so compelling. SDN is among the most over-used and mis-understood topics in tech, most often because vendors struggle to explain why the technology is relevant to their customer's use cases. Ravello doesn't suffer that problem, though.

Ravello's SDN enables the forklifting of your on-prem VMs through an importing process. But when your VM is uploaded to Ravello, how does it communicate with your other VMs? Through an SDN solution, that's how. You can either roll your own virtual network, or use Automatic Network Generation and Auto Binding to have Ravello do the work for you. This is insanely useful if you're less interested in networking and more interested in testing your application.

Basically, Ravello went into hiding after the Oracle acquisition, and now we are getting a peek into what they're been up to. Cool tech and practical use cases: a rare combination these days!

Note: Many thanks to John Troyer for setting up #rbd1, and for also setting up a post-event personal demo with Simon Law from Oracle.