Go get 'em, Lawyers!
Popcorn at the ready.
Microsoft has revoked the Most Valuable Professional status of an Australian developer who publicly complained that the programme was effectively asking members to spread marketing material. The software programmer, Geoffrey Huntley, took exception to Microsoft’s efforts last week in the following tweet. I see the microsoft # …
What you see here, is a lot of effort, money and drive being spent on trying to discredit their competition rather than focus on their own innovation. Microsoft is trying to make Azure the only Cloud choice for Windows customers through licencing changes that penalize their own customers who want to run on other clouds. They would be better focusing their effort on making Azure reliable and innovative.
The question is do these tactics work in the long run. They need to be careful that it doesn't come back and bite them.
Microsoft are annoying about Azure. Everything they do is Azure.
Not everyone wants to be or should be tethered to the Cloud and it only really benefits Microsoft.
SQL Server is good software but a lot of us like on-premises hardware & software.
And what about certifications? As far as I can see, all of the SQL Server certifications have been discontinued and only Azure ones have been newly announced.
Normally Microsoft would introduced the next edition of SQL Server by now.
I wonder how long SQL Server has to live before Microsoft tells us that we will all have to move to Azure.
Just like Oracle and IBM with their clouds. If you want to buy any software stack from them, they'll push you to use their cloud offering rather than on-premises because it'll be cheaper/easier/better/whatever.
Google and AWS have a different tack, since they don't sell on-premises software stacks, so the sales pitches are different.
"SQL Server is good software but a lot of us like on-premises hardware & software."
To add a real-world example, my employer's factory is controlled by several computers across the floor, centrally controlled by a SQL Server database. The db server is housed in the same building (though backed up off-site). I can't think of one good reason why this database should be moved to the cloud, and one shining good reason why it shouldn't: If the internet connection goes down, we don't want the factory to grind to a halt. Anything likely to take the server down on-site would probably take a good chunk of the factory down too.
Answers, in order...
1) We have multiple internet connections to the building. The site is fairly rural though, so if one goes down then the other stands a fair chance of going down too due to shared infrastructure upstream. Regardless, the factory carries on ticking.
2) I am not the administrator of that particular machine so I don't know the fine detail, but I am under the impression that there is a standby machine to fail over to. Current uptime of the machine is measured in years, it was last taken offline by an unannounced power cut that also cut power to the factory (out of hours, thankfully).
3) My original comment stated there was an offsite backup.
In a nutshell, the risk from losing service to a cloud provider is judged to be much greater than losing service to our own machine. Current uptime figures for the machine and the internet connections bear that out.
“1) We have multiple internet connections to the building. The site is fairly rural though, so if one goes down then the other stands a fair chance of going down too due to shared infrastructure upstream. Regardless, the factory carries on ticking.”
I’ve worked for organisations with “secure facilities” in rural areas miles off main roads and have procured diverse data circuits along with diverse power to ensure their ongoing operations, this was 20 years ago.
There are now many more options for diversity and better encryption if needed. Depending on your latency requirements satellite is even possible, 3G/4g/5g, trusty microwave links and other wireless technologies are available.
The machine has been up for years with no patching, that is not a badge of honour.
If it’s important enough, spend the money on proper availability solutions. Network connectivity isn’t just internet connections.
You say there is an offsite backup which I assume is tape or disk which is good, has the restore been tested?
Rather than continue to argue over the details of the specific box in question, maybe limit the discussion to the original point that was being made? To whit, there are legitimate business reasons for keeping systems on-site.
I agree there are many and varied ways of ensuring continued internet access to a facility, but you're crossing the point where you're throwing every technology you can at this to prove it can be done rather than consider whether it's viable. There are some cases where the cost / benefit / risk consideration does not bear out.
You make a stupid assumption about their factories setup, and when he/she politely clarifies, your response is to double down with an even more ridiculous post.
La de da, so you've worked with diverse circuits. We are all familiar with those, and I'm sure when the OP said it was out of his control, he knows the setup of his location better than you.
You then suggest satellite/mobile and all sorts of stuff for him to fix a problem he hadn't got.
Next you assume his network is crap, although nothing has been said to suggest it is.
Final dig, criticise his backups, again with no evidence.
All to try to force him into using an impractical cloud solution.
Next time you're wrong, admit you're wrong. Your arrogance in assuming you know more about his situation than he does is breathtaking. In fact, I assume you're a teenager who "knows everything", easily influenced by "cool" buzzwords, who's never actually worked out in the field.
Not cool at all.
"I’ve worked for organisations with “secure facilities” in rural areas miles off main roads and have procured diverse data circuits along with diverse power to ensure their ongoing operations,"
Very nice. Not very useful though. You can do lots of things to improve a network connection. All are expensive and in this case, none are needed. The server concerned is required for operations inside the building. It is not required for operations outside the building. Why should they spend on lots of network links just to show they can put the machine outside the building anyway?
If you want, you can buy a refrigerated vehicle and hire a full-time driver for it, just so they can go retrieve chilled food and bring it back to you. Or, you can have a refrigerator in the kitchen. If the only person eating the food is you, it's a lot cheaper and faster to chill the food in the kitchen. That doesn't make the vehicle idea bad in all cases--you might operate a business where you have to bring chilled food to lots of different people. Still, you probably don't own such a vehicle and you have a good reason not to.
The weak spot we all need to be wary of is the accounts department. Buying and upgrading servers and software packages gets dumped under capital expenses and goes under the 'bad' column, as it can have large one-off costs. The 'good' column contains operating expenses, which are relatively fixed known amounts per month, such as cloud services.
Once your accountants get control, none of your logic about owning on premise equipment will ever sway them.
Idiotic suggestion. Add cost for nebulous benefit over current solution.
And what about those times when Microsoft cannot authenticate over the internet?
I can't be the only one whose enterprise bought the MS spiel about cloud-based log-on, then couldn't get email or launch chrome for a day when heir server farm was "stressed"
I'm with the factory. Buy a second server if not there already and cluster-up. Don't forget your UPS.
Because I've sat through the 2004 blackout *and* the time some twillup cut a cable in the Fort McHenry Tunnel and took out internet for everyone northeast of Baltimore.
"you could just add more internet connections."
Oh come on. The stated use case is about as clear a don't-use-cloud situation as you could imagine short of an airgapped environment. It's not cheap to run extra internet connections which you intend to be redundant. Cable connections may use common infrastructure, so you either have to pay for installation of alternate paths or hope that an issue with one won't bring down the other. Fixed wireless connections may not be available depending on the size of the factory, are prone to congestion, and may use common infrastructure as well. Satellite might be the best alternative to avoid those problems, but that also depends on the weather and available satellites. Meanwhile, from the sound of it, the server doesn't do anything for people outside the factory, so it's a lot more important that it is available to the other things in the factory than to the outside world.
There are at times advantages, sometimes significant ones, to using the cloud. However, even if the cloud providers manage to improve their uptime to 100% and reduce their prices by an order of magnitude, there will be some cases where it's still not the right decision. A situation where the users and the cloud are separated by unreliable or limited network connections is one of those.
I'm sure that the cloud-deniers have flagged me as a cloud Kool-aid drinker, but you are playing the fool.
I learned SRE at Google. I know EXACTLY what it takes to provide >5 9s. I know what it takes to provide 4 9s as well.
And while I've not said it before, the factory floor is the kind of place where both 5 9s are likely to not be enough AND where it is not all that hard to do better on prem.
The trick is that if power to the factory goes out, the computer doesn't matter. At all. That's actually an amazing amount of "free" reliability. In fact, while I have stated here that I consider 6 9s to be theoretical, that was only the context of providing services to the internet. Not having to worry about connecting to the internet, or what to do if an electrical substation goes out--normal SRE just doesn't get that kind of luxury.
So, set up three servers (yes, I AM an SRE :D), in different parts of the factory. Each with it's own UPS, each capable of running the factory by itself, with proper failovers in place. Things might get a little interesting if your factory is running 24/7, but if your code uses feature flags, it should be NBD. Profit.
(BTW, if you are curious why three is the minimum, it is to permit one planned & one unplanned outage at the same time. Hardware does need to be replaced from time to time.)
so the problem with failover is that it is not easy to determine when to fail over. who watches the watcher.. a common failure mode i saw was when two storage units mutually declared the other dead, and both operated as master. that was the source of many bad days. assuming that you require durability in the database, you're going to need a paxos-like algorithm to solve this problem. i've got a hunch that no traditional sql server provides this.
Unfortunately, it is not that simple.
The cloud host may go down.
The supposed answer to that is to have each internet connection connect to a distinct Point of Presence. But then, there is the assumption that each PoP has the same version of your database. Remember that PoP's don't magically all have your data at the same instant. It has to be replicated. The user has no control over that process: how often it is done, or the technique used, so you may be served data that is chronologically out-of-date.
Marketdroids have been a form of artificial intelligence for a while - you don't really find natural intelligence among them - nor anything "natural", for the matter. Everything in marketing is artificial - their language too.
They are all part of a hive which fully conditions their thinking and acts. "Social" sites have amplified that hive - and they need to turn everyone into their drones.
This became real for me when I was required to take a buisness writing course. For one exercise, we had to respond to somebody we were giving bad news. I had marks deducted for NOT using the passive voice.
This is merely a microcausm of the things that disgusted me about that course. There were many times that we were given advice which, in by subsequent role as a technical writer, would have gotten me into another round of editing.
That is a term that is starting to join the sewer of "socialite".
Gandhi was an influencer. Thomas Edison was an influencer. Genghis Kahn was one hell of an influencer. Alexander the Great was the first influencer.
All these people have is a Twitter account. What do they actually accomplish ? Nothing.
There was an Austrian chap in the 1930s/40s who was an influencer too, can't think of his name right now, Adenoid something? I think he disappeared into obscurity, probably fell foul of Twitter's code of conduct or whatever. Last heard of when he contested the North Meinhead, sorry, Minehead, by-election in the late 60s...
Good to see such integrity - and had to believe that a company would require people who aren't even staff to shill for them.
Even working for a household-name company (hence AC) while I'm happy to talk about the good stuff the company makes I have no desire to run down the competition - they also make good stuff and when they have better products it is just motivation to get ahead again.
They have been doing this for decades. Research on how Microsoft got their MS Office Open Document Format an ISO standard for just one example. Hint: they asked partners to become voting members of ISO around the world, funded them and provided statements they should use in the meetings.
Microsoft is 50% marketing, 20% software development, 30% bullshit.
This post has been deleted by its author
It does. It's just one of those acronyms where context is required to avoid (or at least minimise) the risk of misinterpretation. Right now my primary interpretation of MVP is also minimum viable product, because that's the only context in which I've been actively using it for the past few months. Before that however, my primary interpretation would have been the sports-influenced Most Valuable/Valued Player, unless I knew I was reading something Microsoft related, thanks to all the years I spent following American Football as a kid.
Sticking with the IT angle, AWS is another one open to differences of interpretation - unless directed by context, my thoughts will initially tend towards a type of railway safety system, and not to a cloudy service operated by Team Bezos.
> It certainly isn't about actually providing the answers to problems posted to the MS Community, from what I can see.
Dear AC, I understand your question is about providing answers to problems posted to the MS Community.
Yes, community answers provide posted questions. Have you read this <irrelevant link>?
Debateable. All that's really happened here is he's been stripped of his MVP status. This might mean he's also lost his ability to post on any MVP-only forums, or to have his posts on open MS forums tagged as coming from a MVP, but as this article so neatly demonstrates, he's not been stripped of his ability to make his views known to the world at large via any of the well known means of global communication not controlled by the numpties at MS.
And it seems likely that, had MS *not* stripped him of his MVP status, then this whole hoo-ha over their attempts to use MVPs as human adbots may well have gone entirely unnoticed, so if their intent really was to try and cancel his message, the end result has been the polar opposite - cf Streisand Effect.
If big tech or media doesn't like your message - there is no debate of ideas, you just get canceled.
If you have a special relationship with an organization, and they don't like your actions or you don't like theirs, then the relationship is at risk. It can also happen if one party believes the value equation has changed: if a sponsored artist or athlete is no longer as visible, they'll reduce their investment. And vice-versa.
This is not new, exciting, surprising, or controversial. It's quid pro quo.
"If big tech or media doesn't like your message - there is no debate of ideas, you just get canceled."
Not true in most cases, and not true this time either.
He was being paid (in Azure credits evidently) to post advertising about Azure. I don't know why; it doesn't seem like a good business decision in the first place. He didn't like that. I'm with him--I would only be willing to post approving comments had I actually compared two options and thought one was significantly better. While I wouldn't mind getting money from the better one for the post, I would probably not take it because it would weaken my credibility by implying I was biased toward them. He complained about the program he was in. Is it that surprising that the people running the program figured he wasn't a good person to have in it? He didn't want to do what they wanted him to do.
Indeed. Whenever I'm searching for help with a MS product and I find a forum thread opened by someone else with the same problem, I can pretty much guarantee that the first half-dozen responses can be skipped over immediately, because they're either all from overly-eager MVPs trying and failing to be helpful by regurgitating what seems to be a common script, or responses to those MVPs from the original question asker, pointing out that their "advice" is just asking them to do exactly the same stuff they'd already explained in detail that they'd tried with no success...
It's usually only once you get past the MVPspam and into the part of the thread where other end users (and possibly also people who'd prefer not to admit to being MVPs) start to give decent advice and suggestions that the thread becomes a useful problem solving resource.
The MVP program has gone to hell. It's obvious at the business function layer what they do: "build or buy" a profiting competitor or new platform, take their staff, brainwash them, evangelize the MVP (Minimally Viable Product), abuse their partners, lock-in customers, overpromise and underdeliver, and walk away with a fortune. Rinse, wash, repeat. Model's never changed, just got more sophisticated in that they have adopted "modern" marketing technologies because they can't afford to kill or buy SFDC. They know Dynamics sucks balls. So they arm their pawns (MVPs) with battlecards, FUD cards, playbooks, etc. and viola, fire up Azure ML, AI, et. al., to help spread the word. M$ is the OG Ransomware shop, from the CEO down. Shady motherf*ckers with too much clout in all forms.
"Now go home and get your shinebox!" is the interpretation a lot of MVPs are feeling.
WITHOUT a multi-hour global all-services outage, then, and ONLY then would it be even remotely sane to consider moving business-critical processes to it.
I really don't understand how anyone who needs more than two and a half nines can even think about using Azure for anything other than a backup of their production solution.
This post has been deleted by its author