Hey we missed a week because I wrote this long thing on how messed up everything is, including security, IoT security, and women in tech; all consciously or unconsciously inspired by our current political season. It depressed even me, so instead of sending it, I shredded it for sprinkling atop future emails.
This week let’s talk about drills and holes — and I promise no more prawnography! But let’s start with my old buddies at VMware. This week we had the big VMware and Amazon announcement that VMware would be running a managed service co-located with AWS all over the world. Everybody immediately started punditing as hard as they could. Read TPM at the Next Platform for an in-depth look (and some cloud economics). AWS and VMware Acquaint As Strange Cloudfellows.
The first take on something like this — and my first take too! — is the skeptical one: if hosted VMware was such a hot market, you’d think that VMware’s vCloud Air and vCloud Air Network (and previous incarnations of its service provider program) would have been bigger successes. (It’s not like there’s no business, just not a huge business so far. Yes, I could upload my Tier 2 and 3 VMs to a number of places, but the VMs aren’t that big and don’t take that much care and feeding, almost by definition.) What makes vSphere in an AWS data center different?
The general punditry also called out IBM and Google as being the Cheated-On Lover and the Swiped-Left Suitor of the deal, respectively. (IBM and VMware announced a deal at VMworld US around IBM offering Cloud Foundation. This new deal is a VMware-led offering with Amazon as its “preferred partner.” Google and VMware have partnered on some joint offerings, but the Big G is nowhere in this deal.)
Microsoft was also painted as the enemy-of-my-enemy who drove VMware and AWS into each others arms, since Azure is a soup-to-nuts hybrid offering that neither of our new buddies could field by themselves. And I’m pretty sure that the entire vCloud Air Network, the set of partners that offer VMware as a service, is a little disgruntled.
Pundits like to see things in black and white. But let’s always try to take the technicolor view.
So first of all, let’s assume that VMware is like a normal multi-billion dollar company and can have business relationships with multiple clouds for different use cases and markets. Everybody is friends if there’s money to be made. Next, let’s assume we’ll get more details as times go on, and let’s assume that VMware and AWS are both smart, and that the offer will be priced right.
So for the bull case, on the question “Does Hosted VMware with AWSaddress a market that Phil’s House of VMs or any other vCloud Air Network provider did not?” For this, let’s look to Gartner — in 2015 AWS was serving 10x the capacity of the next 14 competitors, combined. This year Gartner waffled back to “many times the aggregate size of all other providers in the market” so maybe not 10x but still huge. The Court will take into evidence that AWS is a sh*tload bigger than everybody else and that yes, simply by partnering with them VMware could shake up the Hosted VMware market.
Let’s get to the tech part — look at Frank Denneman’s post and notice the part on elastic scaling, then sit with that and think for a minute. What if you could spin up and down ESXi hosts as easy as you can spin up VMs? For maintenance, for capacity, for test/dev, for CI/CD, for any reason at all? I’m no expert, but I don’t believe any other hosted VMware offering has that kind of integrated capability — Rackspace and IBM Softlayer both have very cool bare metal services, but they’re a little more DIY than this upcoming offering direct from VMware. And in this cloud market DIY is Bad — before long you’re compiling your own Gentoo and trying to get Cinder running. DIY is a solution for lunatics, not businesses.
So did you you sit with that elastic scaling bit above? Interesting, right? Maybe even game changing? No? Not quite? It certainly could change the way you use VMware vSphere, perhaps even quite disruptively. Now add AWS services — databases and data warehouses, load balances, Lambda, and more. All in your vCenter, just a click or API call away, and in the next rack over. Interested yet?
And the final thought for the bull case. If VMware really believes that the future is hybrid/multi, then it must create a future where multi-cloud is ubiquitous and easy. It must also create examples and convenient hybrid platforms that have VMware in a central role and give people a damn good reason to use NSX. It seems like IBM (plus future Cloud Foundation partners) and VMware on AWS would be good role models for a future hybrid- and multi-cloud world? Nicht wahr?
Which brings us to drills and bits. There’s this B-school saying (Haaah-vahd Business School to be exact — Theodore Levitt as quoted by Clayton Christensen), “People don’t want to buy a quarter-inch drill, they want a quarter-inch hole,” meaning don’t sell the features of some product, sell the benefits, because that speaks to what people really are looking for. (Of course, nobody really wants a hole either, they really want some shelves. Well, not shelves really, but they want to get the junk off the garage floor. Well, a clean garage is nice, but really their motivation is to stop their spouse from mentioning the messy garage every time they park in the driveway. But I’m sure the need for a hole is in there somewhere. But the techies keep arguing about 12V vs 20V cordless drills.)
For instance, I think sometimes we have that drill/hole thing going on with hyperconverged infrastructure — nobody really gives a crap about the hyperconverged part, they care about the super-simplicity and power they’re getting out of these new platforms. It just so happens that converging compute and storage was a great way of getting that simplicity over the last five years. We are starting to see that the next gen of storage platforms are arguably as simple, and with NVMe over Fabrics the form factor arguably doesn’t matter. The biz folks know that apps are fast, and only the nerds are arguing over the drill bits — i.e., whether the disks are in the same case or next door.
What if AWS was a drill bit, and agility was the hole? Have we confused the bit for the hole? And what if you could now get that an AWS-like pay-by-the-drink service with VMs, containers, and vSphere hosts directly from VMware, and enterprise-grade beside? Is your garage clean enough yet?
Read these comments on Hacker News talking about why they still love and hate their data centers. Think about drills and bits.
I’m a huge believer in colocation/on-prem in the post-Kubernetes era.
Meanwhile, Google will embed its SREs (Site Reliability Engineers) into your teams. This is a good way to put your DevOp where your mouth is. Here’s Google’s Plan for Calming Enterprise Cloud Anxiety by Yevgeny Sverdlik at Data Center Knowledge.
Here’s one worth a click that I was going to include last week. The Real Reason Women Quit Tech (and How to Address It) by Rachel Thomas on Medium.
The first step, and the most important step, to improving diversity is to make sure that you are treating the women and people of color who already work at your company very well.
Read the whole thing, as they say, as well as the links, to find out why your company is not doing as well as you might think or want with respect to actually treating the people who work with you right now as well as you think you are. Unconscious bias is an insidious thing.
From the mailbag:
- Chris Evans on Kitchen Prawn: Always ensure your spashbacks are made with tempered glass. Had to replace ours this week after we had a heat-generated crack go across the old one.
- Bill Petro: The data center is dead, long live the data center. There is a strangely “moral” judgement going on in the industry. Kind of like the recent election campaigns. Data centers are bad if customers build them, but they’re good if public cloud companies build them: AWS, Azure, Google Cloud Platform. They’re not the sexy place to be. For example, Wall Street beat up Cisco recently in their annual earnings call on the flatness of data center sales. IOT, Security, and Cloud are “growing” while DC is not. However, DC still several orders of magnitude larger in actual revenue despite the flatline. (Hey, Bill is on the market — startups, if you’re serious about a well-trained field, see Bill!)
Listen to Geek Whisperers #122: Getting Certified with Gregg Robertson, VCDX. Gregg takes us through his process for getting certified, which involves both public accountability and a lot of nights and weekends. (Since recording this episode at VMworld, Gregg has succumbed to the Geek Whisperers Curse and changed jobs, moving to Dell EMC. Not our fault!)
I’ll be headed up to CloudNativeCon November 8–9 in Seattle. Please come be my Kubernetes friend if you’re there too.