I've seen "good agile" and I've seen agile twisted into a hellscape of management torture. As an organization tool, agile is 'fine' - but it doesn't solve for a broken organization. If you can't or won't manage expectations, if you can't or won't prioritize, and especially if you can't or won't do real product design, then nothing is going to solve for that. When you have management evaluations being done on checkin count or LOC metrics, then there's nothing really agile will do to save you. When your product organization gives you 30 #1 priorities and wants them all at the same time with a team of six, then you're screwed no matter what process you follow. Sure, you can always make it worse - add SAFe on top of agile for instance, and watch the gears grind to a slow gummed-up mess of cross priorities and endless quarterly 'iteration' meetings where everything has to come in at the same time and releases can't be done independently because nobody wants to put the effort into making that happen. *deep breath* Where was I? Agile... It's not the process you hate, it's how you're doing it. Software development is hard, no matter what, and agile was a process to take some of the inter-dependencies out of the mix and decouple and manage expectations by providing buy-in and shared context. If you're not doing those things *at the least*, then agile can't save you because you broke it.

The comments this time are spot-on and hilarious. Gartner is certainly smoking the lucrative billable hour crack. The hardest part of a migration is moving the data, and... that's about it, no matter what the destination is. Yeah, if you have a lot of them it will take time. A lot of coffee breaks mostly watching bits move. And yes there will be special edge cases that require some research. But if all you're doing is VM's and storage, it's really not hard.

Broadcom either was counting on the FUD of doing migrations, or they just knew they would make more money regardless. (Both probably). The competing hypervisors are rushing now for some level of feature parity, and it's not that high a bar to clear for most use cases. And Xen, Qemu, KVM hypervisors are more modern than VMWare, and handle more diverse hardware. Right now as I see it the worst choice is Microsoft, with their flaky commitments to virtualization and frankly poor I/O compared to a Linux based hypervisor.

Known VPN services have identifiable server addresses that can be blocked. Instead, you can set up a cheap raspberry pi (or other) at your home and use an encrypted SSH connection to that [raspberry pi] from far away. Then turn on your SOCKS proxy (part of WiFi Details on Macintosh) and check to see that your IP address shows to the world you access as that of your raspberry pi. I do this all the time, including right now. It also helps to watch sports events.

... It's how you use it.

On the plus side, there will be a whole lot of nice lightly used GPU servers for sale in a few years. Mostly these machines have moved away from the PCIe card format that consumer GPU's use, so it will be harder to sell them off individually.

Honestly, you can roll your own, and that's great for a tinkerer and someone who manages their own infrastructure like a home lab or the like. It works, it can be cheaper, and you get what you spec. Maybe it's too easy actually, since you can screw it up and if you have the right backups then it's time wasted setting it straight, and if you didn't have the backups, then WTF get your head on straight.
You can buy something pre-rolled. Synology has great software, and while I think the hardware is a bit underwhelming, it gets the job done unless you're trying to run your whole setup on the NAS which even tho the software can do, it's gonna fall on it's ass with a bunch of containers etc running. QNAP, while I don't have the experience, can do it too. Synology or similar would be my first stop for someone who's looking for something to get going with and wants it stable and maybe it can do a little more besides.
Beyond that tho, like most things, you will shift the bottlenecks around in your setup - if previously you needed more space, well now you got a ton, only it's not quite as fast as you were hoping. Honestly once you got backups sorted and a good reliable platform, the next frontier is networking - 1GbE is just not fast enough, IMO, for modern arrays. 2.5GbE should be the minimum and 10GbE is really where you should aim, and really it's not that much more expensive anymore. But of course this means you might need new cables, a new NIC, a good router or switch, maybe an add-on card for the NAS, etc. But you should consider network before you consider other options, like caching or memory upgrades, etc. Seriously, it makes a huge difference. Unless you're on wifi - then you're stuck until wifi 7 is common.
Good luck!

Thanks.

Gee while you're shooting for the moon, why not claim 100% of the Windows market, and make it THIS year! Cause that's just about as likely to happen. The only possible path I see to even a small percentage, and it's really a longshot, would be if MS decides to ditch the "legacy" architecture and pull an Apple with emulators, etc. But MS is not Apple, and doesn't control the hardware platforms, and the Intel/AMD architecture isn't going away anytime soon, especially if Intel and AMD have any say in it.

My personal experiences with Fedora have been good over the last 10 years or so, with occasional hiccups. Today I run it mostly in VM's, and 39 has been super stable. I tried the 40 Beta a few weeks ago and felt it wasn't quite ready yet for my use-case, so I'll wait a bit before trying the release, and maybe in-place upgrade a VM that isn't critical. Bleeding edge, like Fedora, sometimes means it can be messy, but I feel they have managed that well in the past releases. I'm sure most of my concern is around the GNOME updates and performance there with VM's, but I didn't try to quantify it due to time and getting-shit-done.

For the most part there's still a few places I would like to see exported from VMS, rather than trying to run VMS on X64...
The first is networked and clustered filesystems - VMS had this down cold. Easy to configure and manage (at least at the user level) and very transparent. Files could be on any node of a cluster. Which brings me to the second:
Distributed Lock Management. I have no idea why this isn't a thing now. We have parts of it, and even SMB has some locking, but it's not the same as being able to generally lock resources that are shared across a cluster. Heck, down to the RMS record level if you want. And that's #3:
RMS. This was the gold standard of record and internal file management. Yes, it's not Unix-like in that files have structure that is known, but damn it made some things really nice to do and removed a lot of bit-bashing and buffer manipulation at the application level. Sure you could just have a "big bag of bits" file that was managed by the app, but you could also have a nice VSAM/ISAM-like file with some defined structure that could be used by some tools.
VAXCluster - goes along with the above - we don't have a good "Pick a node, any node, and run there" concept AFAIK. Linux clusters are not that. Maybe Kubernetes and microservices have made that obsolete, but we had it back in 1984 with clusters.

And before I get a lot of replies - yes I'm aware there are solutions for all of these in some form or another, but nothing I know of gives the whole package and available as a resource to just use.

Someone (Not me, no time) needs to write a DCL shell for Windows. DCL was a great scripting language. There was once a commercial shell available (maybe for DOS?) but I've never used it.

Three years was my guess too - Google doesn't have a good track record with non-advertising related products. I'm actually surprised that Drive and their office suite is still around except for the fact they use it internally and it saves them a ton on Microsoft licensing and was a driver for Chromebooks etc.

Wondering if there's some reason (other than infrastructure) why there aren't multiple charging "channels" - the trend now if higher voltages to charge faster, but wouldn't some kind of parallel set of circuits works as well, even at lower individual voltages? Divide up the battery pack into charging zones and charge each at the same time. Yes, more complicated charging circuitry but really I think it's do-able.

Nice Troll. Have you read her "publication record" (which for an academic is called a C.V. btw)? "Much of it plagiarized" is a far cry from some unclear citations, and "not worth plagiarizing" is your academic appraisal, is it? Really, she has one big job, and that's bringing in money and addressing donors, and all this noise makes that hard to do. At Harvard academic concerns (and much of the policy too) are managed by the different schools themselves, and don't have much to do with her or her office. Yes, she's the spokesperson, and again because of the distractions and such she made this decision. Harvard's so-called prestige is untouched and it remains one of the most rigorous and well respected academic institutions. And yes, I'm a Harvard grad myself, and believe me I hold them to high standards, as I suspect most alum do.
So, whatever right-wing "anti-elite" BS axe you're swinging can be put right back up your arse where it originated.

Not necessarily incoming fire - but I'll try and be nuanced. Not all "formal education" is created equal, and not all self-taught are lacking in understanding. I've interviewed hundreds of people for jobs, and it's actually hard to predict success from any documentation or training program. Some things are universal tho - if you have an MIT engineering degree you're more likely going to be more rounded and versed in the fundamentals than someone from a community college with A+ or something certs - but not necessarily! I have met many "masters of CS" that couldn't tell you about caching or hashing for example. And plenty of people with Java certifications that don't know about multi-threaded data safety, and sometimes not even about locking. So i guess I sort of agree with the OP, and sort of don't agree. People with certs may or may not be better candidates, but in my experience it doesn't matter if they have a cert or not as it's not a predictor.

"You must work harder to increase my wealth!" He is right on the other point, in that any increase in productivity would just be met with increased productivity of corruption. But basically this is a privileged person crying in his beer that he doesn't make the most money, and really doesn't give a crap about whose face he has to step on to get more (or put another way, doesn't really care about your life balance.)