Stack Exchange made an announcement with much ado about how they moved to “the cloud” after having a landmark deal with Google earlier in the year. You can safely assume they moved to GCP which is whatever, if it works best for them, awesome. However, I don’t really see this as a good thing or a win and I found the overall post to be… written like a madman who has no idea what is reality or how anything actually works, as in logic seems to be a foreign concept, let me explain.
The main section starts with the heading of, “Why are we moving to the cloud?” and this is where they opine about the many reasons it makes sense to move to the cloud. This is also where the whole post falls apart. It starts out by pointing out how their current setup (on-prem) is extremely scalable and cost efficient to which immediately after stating this proceed to talk about how it cost too much in terms of time, energy, and money to keep things working on-prem. Now, I find that a bit odd to start the whole section off by stating that everything is scalable and cost efficient, only to then dunk on how it wasn’t. So which is it?
Eventually the reader will find out that the data-center they were at was closing, which was the impetus for, “something new” (aka cloud). I found it interesting that they didn’t say better, more efficient, cost saving opportunity, or any number of things. No, it was “something new”. We’re given, by the writer of this post, the rationale that we’re leaving a highly scalable and cost efficient architecture to do, “something new”. Well, if that’s all it takes to burn down Rome and start again then I guess we should just start something new… oh look, V1 shipped of Generic Software product, now I guess we shouldn’t do anything with it and just do “something new”. Awesome logic. Bravo. Such great thought leadership. Gods among men.
At the last bit of the main introduction you find out that they reverse everything they’ve said, “To make it explicit, however, we are not going to the cloud to save money. We know that cloud is often more expensive than running your own hardware.” and ” However, the cost is worth the new flexibility. Instead of projects being delayed while new hardware was procured, installed, and configured, we can spin up new capacity in minutes.”. The argument here is that people can’t project what they need a couple weeks in advance for a project and configuring something is just too hard, so let’s all shoot from the hip and run up a bill because no one can plan or knows how anything works. Excellent. Real 10x thinking.
This post is everything wrong with cloud migrations. Do it if it makes sense. Don’t do it if it’s dumb. People in technology are so beholden to “something new” and blinded by actual great solutions. The entire post honestly reads like an intern in Technology wrote it for their end of year essay… they have no idea how anything works, they don’t care, they just want to play with new toy. Everything is surface level nitpicking while completely massaging over all of the negatives in order to do something different, or as they have said, “something new”.
I don’t know what happened to Stack Exchange, but it seems they have no one there who understands technology or the world it encompasses which is a far cry from where they were when I started answering on the platform. This along with their hyped AI push is truly the end of an era. I’m sad to see it continually get worse and worse, but to quote them, “use this as an opportunity to explore something new?”.
I’ve had several clients that were wooed by really, really attractive financial terms from Google. Sounds like that’s what happened to Stack as well, given this line: https://meta.stackexchange.com/questions/404231/we-re-finally-going-to-the-cloud#comment1347891_404235
“I understand this is an interesting question but as you guessed it gets a bit too close to financial details that I can’t talk about.” Wouter de Kort
Absolutely, especially given they sold all their data to Google for AI training. My guess was it wasn’t solely financial based but probably involved free compute. It’s still sad how they write their public facing blogs, though. The place has sadly been going downhill since the people who actually care about it and used to make a difference have mostly left. I still try to help where I can, but the platform (with the new CEOs and such) is just moving in the wrong direction.
Are you crying because they didn’t choose Azure?
If they chose GCP just for money, then Azure should have pushed for a good offer. If Azure can’t match it, then it’s their choice, it may be wrong or it may be right, time will tell, and you can be whining as you want, but that doesn’t mean that you can claim they have no idea of what are they doing, or that they are dumb.
The “AI hype” might be just that, and most technological companies are driver by hype, but it could also be the next big thing and StackOverflow doesn’t want to get behind, it’s just a bet that could work well or bad, but here seems the real “intern” in Technology is someone that does get it.
Hello Anonymous Intern! I see you’ve entirely missed the point but continue to tow the company line without regard. You’ll make a fine fully-fledged sheep one day, congrats!
“Stack Overflow has existed since 2008. At that time…. We built a monolithic application that scaled incredibly well and we squeezed the most out of the hardware we had.”
I don’t believe the post meant to claim that the monolithic application has *always* been incredibly scalable or that there are not current scalability pains. When you are a small to medium sized business, the constraints of using on-prem datacenters can cause a huge amount of engineering compromises due to having to self-support everything you build. It can be cost effective, but limits flexibility in terms of scaling from where they are now. Add that into the fact that their entire business model is being forced to change in order for the company to survive as a for-profit organization, and I think this entirely makes sense.
But I also don’t think Stackers would feel right about just calling the current/old architecture trash or saying it wasn’t scalable at all. It did serve them well for a long time. I don’t think they should necessarily have to dig out all the skeletons from that closet and hold them up for the world to see.
Anyway, I’m really only commenting because you wrote that “it seems they have no one there who understands technology or the world it encompasses.” As a former Stacker, I can promise they they DO have many employees who are super smart and understand the various tradeoffs between on-prem and cloud technologies. There is a lot of romanticism of the “old stack overflow” model and people seem eager to believe that all the intelligence has left the company, but there are still many great, hardworking, knowledgable engineers working there. I look up to many of them.
> Add that into the fact that their entire business model is being forced to change in order for the company to survive as a for-profit organization, and I think this entirely makes sense.
I have no doubt they are getting in-kind of a sweetheart deal, their post, however, glosses over this completely and goes out of its way (as I quoted) to state this is not the case. It very well still could be, but then why not say it? Seems like it’d be a nice win. /shrug
>Anyway, I’m really only commenting because you wrote that “it seems they have no one there who understands technology or the world it encompasses.” As a former Stacker, I can promise they they DO have many employees who are super smart and understand the various tradeoffs between on-prem and cloud technologies.
It doesn’t come across that way or seem like it. For example, they point out that it was just the end of the world to have someone go to the data center. Now, they get to wait for days/weeks/whenever (maybe they will get super top of the line Google support dedicated people?) to deal with their submitted tickets on things… and before anyone calls out the ridiculousness of that, I have very intimate first hand knowledge of exactly how that goes (and I _know_ for a fact you do as well). None of this, of course, was called out in the blog post, why would it? Everything new is sexy and awesome, everything old sucks and is broken, which is indicative of people who don’t actually understand things.
Loved hearing from you!
Cloud has a lot of benefits – Economies of Scale, lower up-front Capex expenditure, additional layers of abstraction via managed services, “serverless”, etc.. Great concepts, but case dependent and certainly not one-size-fits-all, I agree!
On another note, I hope you’re doing well these days! You were looking jacked as hell last I saw. Keep up the good work!
Hey Dave! Cloud does have some benefits as you’ve mentioned and it’s great when that’s what is needed. It also has many drawbacks that most companies are really having a hard time with these days.. I’m still of the pick what’s best for your specific needs. In the day of “consume more hyped product X” it’s unfortunate that people aren’t making those specific needs decisions anymore (well, at most places it’s a top down push from whomever who doesn’t actually know technology). It’s all cyclical, we’re on the apex/backswing of the cycle from what I can see… only time will tell!
Great to hear from you again! Thanks!
I second the decision, and additionally suggest them to abandon the outdated, long-dead Microsoft SQL Server in favor of Cloud Spanner.
I don’t know how much Google is paying you for sales, but I assume your quota is looking a little empty these days. Sorry to hear :'(
Awesome, now I can NOT rely on Stack anymore from the cloud for good clean answers because of their asinine reputation algo. I deleted my account years ago after realizing there was no way to get beyond a beginners rep and often you could not even edit your own posts. What a joke it became. Now flying even lower in the cloud. I hope they enjoy the dreaded “Vendor Lock-in” from the 1970s
To their points:
1) “We know that cloud is often more expensive than running your own hardware.” – Yep Especially SQL Server
Putting your server to sleep to save money is a red herring. IOPS anyone?
2) “However, the cost is worth the new flexibility.” No, its not. Thats MIS employee bonus money for staying on-prem
3) ” Instead of projects being delayed while new hardware was procured, installed, and configured, we can spin up new capacity in minutes.” – We can too with our Spare capacity on -prem – duh
4) AI is “quick solutions for the lazy” but thats another discussion