VMworld 2016 Keynote

Unfortunately not attending VMworld this year. Decided that there was probably not enough new stuff to justify the trip, and am planning to attend Ignite next month instead.

Looks like I made the right call, because I just watched the day one keynote, and I found it very underwhelming.

The talk Pat gave about the rate of cloud adoption is very similar to the numbers I calculated in my last blog post. When people make wild predictions on the rate of cloud adoption they forget three things. First is how many legacy workloads actually exist, second is the fact that workload numbers continue to grow, and third is the ability of cloud companies to accelerate the data centre build rate. It seems to me that rather than increasing their data centre build rate major cloud vendors are starting to slow down the rate of expansion as they achieve broad geographic coverage.

The problem with presenting a long term cloud adoption strategy is that no matter how true, it makes you sound like you are not looking forward. It does not matter how long the old technology hangs around for, nobody is investing in legacy. Everything needs to focus on the cloud if you want to maintain credibility, you have to put it all on the line. Microsoft is a fantastic example of how to do this correctly, they are driving everything to cloud, while still releasing updates and making the majority of their revenue on legacy platforms. For Microsoft it does not matter how long it takes, they are making money every step of the way. You would never hear Satya talking about a fifteen year transition plan!

I thought the demos and customer stories were weak.

The AWS NSX demo has not advanced much since last year. It is relatively easy to do micro segmentation; that can all be done on a virtual gateway. What I want to see is distributed logical routing. I am guessing that is much more difficult, since they don’t have access to the hypervisor.

The first demo scenario showed that AWS admins can't configure cloud networking poorly. The VMware answer is to layer another technology on top of AWS to do service discovery, monitoring and cost analysis, and to enable firewalls and encryption. The problem is that most of this functionality already exists natively in AWS and Azure. I am struggling to understand the VMware value add.

For example you can do in transit encryption with AWS using SSL and AWS already has a virtual firewall with the WAF service. Networking in AWS and Azure requires a lot more integration than just virtual switches on IaaS virtual machines. For example in AWS what happens when you use ELB and autoscaling, does the VMware stuff still work? What about if you are utilizing non EC2 services such as RDS and Lambda? Are they visible to NSX? There was nothing here that could not be achieved with native AWS tools, and I am guessing that it would be easier and cheaper to fix your AWS network configuration than add another layer. Hopefully there is a lot more to this.

The example on moving databases servers for the IOT example was also nonsense. Nobody uses VMs for databases in AWS, they use RDS, which has native cross region replication functionality. Also AWS has a bunch of stuff to solve the global latency issue, from CloudFront content distribution, which can happily do writes as well as reads, through to latency based rules in Route53.

The problems that VMware are trying to solve only exist when customers lift and shift their on premise system directly to AWS, and that is not the normal scenario. Most customers do some refactoring to migrate apps into the cloud, at least at the database and web layers.

I like the fact that VMware is focused on these solutions, and there is significant potential in being able to deploy a true cross cloud overlay network, but so far the critical pieces are missing. We need logical routing, VXLan spanning and L2 bridging across on premise and public cloud. That is what will enable cloud migration, disaster recovery, simple provisioning and automated bursting scenarios.

Maybe next year? After all, there is no hurry, right?