Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On
-
@scottalanmiller said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
And I think that VMware, through many market changes, is moving more and more into the "small, but better deployment space." Fewer deployments to maintain, but those that remain are better, and I'm sure pay more. Not entirely unlike Microsoft moving customers from perpetual licenses to O365 - it actually decreased their market penetration, by a lot, but it increased revenue and decreased cost. Big wins, but market share went down.
That's where I see Vmware. Market share is shrinking, it's not the go to product any more. But better customers, at higher revenue. That's better for Vmware.I think though that on-premises workload while not shrinking are not keeping pace with cloud-hosted workloads. Now plenty of those workloads end up on clouds running vSphere, but even those that do not can still end up managed by VMware. VMware is more than vSphere.
NSX-T/VeloCloud runs just fine on Public Cloud, Containers, even KVM etc. I've seen iSCSI from vSAN shared to bare metal Oracle RAC clusters. Airwatch (leading MDM platform) has really nothing to do with vSphere. WaveFront at purchase couldn't even inject metrics from vSphere and it was a while before they added it (It's focused on application telemetry and ML of that datasets). VMware Horizon View can run on Azure, and the CMP products can manage Azure/AWS etc also. CloudHealth provides compliance across all public clouds also. With the Outpost announcement, I will be able to run vSphere, on AWS leased hardware that's installed in my own datacenter and consume EBS volumes into vSAN while layering AWS RDS on top to provide Postgres or Oracle databases as part of a blueprint to a Project Tango application stack for the ultimate multi-vendor meta sandwich... Spending over a billion on R&D, and a few billion on M&A gets you some damn nice toys
-
@StorageNinja said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
Now if your argument is that you are secret squirrels who:
- Are committed to never purchasing anything from a company or distribution.
Of course, if using anything open source is considered "secret squirrel." But this is how IT buys loads and loads of stuff. "Buying" isn't from vendors or through channels. It's free acquisition. Turning to a channel for the majority of things that we use seems weird. We use them when we need things from the channel, but so much of what we use isn't from a channel.
-
@StorageNinja said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
- obfuscating downloads and compiling all security updates from source to avoid incrementing the download counters.
I get this, but there are SO MANY download counters. We know that they are not unified. Even accidentally people will do this. Whether it is that we use a cloud that caches those things, or we just use a cache, or we use our own tools. Having worked in the enterprise, everything like that was run internally. Tens of thousands of installs, only one copy would have been known about and it was a complete download that would be discarded in any stats because it doesn't represent anything. Download counters are interesting, but don't even start to give a complete picture in any way. And so many of them don't report to any central source. There is simply no "source of truth" on these things.
-
@StorageNinja said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
- Disable all phone home, and kill any support telemetry features.
While loads and loads of shops do do this, so many things that we use simply don't do this. It's a mechanism that is rarely deployed.
-
@StorageNinja said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
- Always using burner emails and phone numbers whose contact information never ends up on one of the bazillion marketing (note all it takes is you forget to uncheck a box once in your career) that exist are not surveyed that might be true.
I don't buy this. Just because my phone number is out there, it's not out there in a way that would make the IDC call me to find something out. If this is how IDC does things, then we know their info is useless.
The problem is, you can't collect this kind of info in a meaningful way. You just can't. VMware knows how many customers it has. MS does too. But RH does not, not even close. They don't even have a reasonable way to guess. And that's just RH, let alone Linux or KVM in general.
-
@scottalanmiller said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
Not entirely unlike Microsoft moving customers from perpetual licenses to O365 - it actually decreased their market penetration, by a lot, but it increased revenue and decreased cost.
Who's costs went down?
-
@Dashrender said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
@scottalanmiller said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
Not entirely unlike Microsoft moving customers from perpetual licenses to O365 - it actually decreased their market penetration, by a lot, but it increased revenue and decreased cost.
Who's costs went down?
The vendors. Reducing the amount of legacy stuff you maintain reduces costs a lot. Maybe even by half in some cases. It is unbelievable how much legacy support costs companies.
-
@scottalanmiller said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
The vendors. Reducing the amount of legacy stuff you maintain reduces costs a lot. Maybe even by half in some cases. It is unbelievable how much legacy support costs companies.
You get fewer support calls/bug fixes, but there's still plenty of CPD costs tied to security on older platforms that are still in the wild.
The benefits of "Cloud first" is you can ship faster. I think we push features into VMC quarterly which is a hell of a lot faster than our old 18 month waterfall and Microsofts 3 year gap on major products. Cloud first CI/CT or CI/CD process reduces QA costs. Now I'd argue Microsoft Windows has screwed this up by thinking the insider program was a suitable replacement for writing tests (It's a huge dumpster fire right now).
-
@StorageNinja said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
@scottalanmiller said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
The vendors. Reducing the amount of legacy stuff you maintain reduces costs a lot. Maybe even by half in some cases. It is unbelievable how much legacy support costs companies.
You get fewer support calls/bug fixes, but there's still plenty of CPD costs tied to security on older platforms that are still in the wild.
The benefits of "Cloud first" is you can ship faster. I think we push features into VMC quarterly which is a hell of a lot faster than our old 18 month waterfall and Microsofts 3 year gap on major products. Cloud first CI/CT or CI/CD process reduces QA costs. Now I'd argue Microsoft Windows has screwed this up by thinking the insider program was a suitable replacement for writing tests (It's a huge dumpster fire right now).
MS' problem is that they are just doing a shitty job with releases right now. It's not related to their schedule or style, it's just bad quality.
The best processes still suck if quality is no good
-
@scottalanmiller said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
@StorageNinja said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
@scottalanmiller said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
The vendors. Reducing the amount of legacy stuff you maintain reduces costs a lot. Maybe even by half in some cases. It is unbelievable how much legacy support costs companies.
You get fewer support calls/bug fixes, but there's still plenty of CPD costs tied to security on older platforms that are still in the wild.
The benefits of "Cloud first" is you can ship faster. I think we push features into VMC quarterly which is a hell of a lot faster than our old 18 month waterfall and Microsofts 3 year gap on major products. Cloud first CI/CT or CI/CD process reduces QA costs. Now I'd argue Microsoft Windows has screwed this up by thinking the insider program was a suitable replacement for writing tests (It's a huge dumpster fire right now).
MS' problem is that they are just doing a shitty job with releases right now. It's not related to their schedule or style, it's just bad quality.
The best processes still suck if quality is no good
The question is - why is the quality so bad? Isn't the process supposed to catch bad quality?
-
@Dashrender said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
@scottalanmiller said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
@StorageNinja said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
@scottalanmiller said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
The vendors. Reducing the amount of legacy stuff you maintain reduces costs a lot. Maybe even by half in some cases. It is unbelievable how much legacy support costs companies.
You get fewer support calls/bug fixes, but there's still plenty of CPD costs tied to security on older platforms that are still in the wild.
The benefits of "Cloud first" is you can ship faster. I think we push features into VMC quarterly which is a hell of a lot faster than our old 18 month waterfall and Microsofts 3 year gap on major products. Cloud first CI/CT or CI/CD process reduces QA costs. Now I'd argue Microsoft Windows has screwed this up by thinking the insider program was a suitable replacement for writing tests (It's a huge dumpster fire right now).
MS' problem is that they are just doing a shitty job with releases right now. It's not related to their schedule or style, it's just bad quality.
The best processes still suck if quality is no good
The question is - why is the quality so bad? Isn't the process supposed to catch bad quality?
Nope, the process has nothing to do with quality. Processes are the excuse, quality is the job. You have lots of processes to play politics instead of just addressing quality.
The "big change" here is the timing of releases. If you think about it, release scheduling has essentially no possible way to directly impact quality. It doesn't affect anything related to quality.
Think of it like fuel efficiency in a car. And you used to check the efficiency once an hour. Now you check it every ten minutes. How often you look at it doesn't change what it is doing.
-
The real question would by "why did my car go from 26 MPG to 12 MPG over night" and the answer might be something simple and obvious.
Like losing your exhaust system. Doesn't seem obvious at first glance that it improves fuel efficiency. But it does.
-
@DustinB3403 said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
The real question would by "why did my car go from 26 MPG to 12 MPG over night" and the answer might be something simple and obvious.
Like losing your exhaust system. Doesn't seem obvious at first glance that it improves fuel efficiency. But it does.
Right. And loosely, checking "more often" means that you can catch things more quickly if something does go wrong. But if you don't bother fixing it when you check, checking does nothing. That's where MS is. They are releasing more often, but they aren't bothering to do the actual job well.
-
@Dashrender said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
The question is - why is the quality so bad? Isn't the process supposed to catch bad quality?
Their process is consider the windows insider group (extreme power users) to be a good enough replacement for proper QE teams, and writing automated build tests.
-
@StorageNinja said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
@Dashrender said in Why I Feel KVM Is the Easiest HyperVisor to Learn the Basics On:
The question is - why is the quality so bad? Isn't the process supposed to catch bad quality?
Their process is consider the windows insider group (extreme power users) to be a good enough replacement for proper QE teams, and writing automated build tests.
Right, the new process isn't to catch bad things, it's actually to see bad things as "not all that bad." Presumably because a shift from viewing their products as being for business to being for entertainment. Remember when Windows 95 was a key tool for businesses, but by Windows 98 they had made sure to put a "for entertainment purposes only" label on the product to make sure no one confused it with something that was intended for business use?
I feel like that's where they are now. At least internally, no one is really thinking of this as a business tool.
