I wonder if large companies get more of a pass with outages then small ones? Bias towards authority in a way.
If a service has a big enough base an outage could almost seem like a pass for everyone. “Oo couldn’t do that this morning GITHUB was down.” Whereas if my small GitLab server was down bosses could be “well why aren’t we on GitHub like everyone else, it’s up now?”
When google calendar was down a bit ago I tried looking at random people responses on twitter and many were “whew I get out of some meetings!”
Perhaps there is some accountability for large scale outages like these but it really feels like consensus is often “shit happens” which is totally reasonable just seems like it could hurt big companies a lot less.
I think an interesting counterpoint to my observation is Firefox claimed they lost a lot of users because Google services just broke in odd ways only on Firefox so users moved to Chrome.[1]
My counter to that would be, even though Firefox was and is a big company it's easy to change. A lot of services like Outlook, Chrome, or Gmail or just downright colossal at this point. Changing from Gmail isn't easy because of all the other services and logins associated with it.
Not to mention, Gsuite, GitHub, Outlook, and such are all very corporate. Higher ups make the decision as to when to use these often enough, so you can't just say "this has kind of failed us a couple times lets try something else".
So all in all, I still stand by my point, entrenched big companies probably get more of a pass.
> Changing from Gmail isn't easy because of all the other services and logins associated with it.
You can keep the Gmail account and start using a new one somewhere else. You start registering to new services with your new account and gradually migrate old services to it if they let you. You end up with gmail for social logins (which hopefully you can migrate too) and little else.
I do have a Gmail account. I use it to login in Google when I work for a customer which use the Google Cloud, when I have to upload videos to YouTube (I logout after that) and for Google Play.
> You start registering to new services with your new account and gradually migrate old services
Try doing this with with thousands of employees with multiple devices and varying computer skills, at the same time, while keeping a business running. Don't forget to factor in multiple 2FA systems, that you also have more service accounts than you probably think, and that your spam filters will change behaviors (for that added bit of uncertainty).
Congratulations, you've identified a market need. If someone builds a tool to do all this (if the market isn't saturated with them already), companies will pay for it. There's plenty of tools to convert to/from Microsoft Active Directory deployments or varying quality, I don't see why Google should be any different.
If a company married Google or any other similar service provider I expect that the migration in case of divorce is going to be expensive. For a single person not so much.
I agree, but I think there's also a bit more to it.
Thinking specifically of cloud infrastructure, if you run your systems on infrastructure that is also used by your largest partners and customers, then your outages will tend to coincide, so (a) nobody cares that you're down because they're down too, and (b) even if they did care, they're busy enough worrying about their own systems that they have less time to come at you.
An example would be a company specializing in Shopify "apps" running their systems on GCP because that's where Shopify runs... if GCP goes down then Shopify itself is also likely down as well, and you'll get more mileage out of the "sorry, GCP is down" justification (if it's even needed).
Yeah that’s a good point, it also shows up in weird ways. Like when Facebook auth went down a slew of unrelated apps did too because they relied on it.
When I was younger I really wanted one thing to do everything, when chat was integrated into gmail is was like “oh boy, one less app!” And then similar feelings when it connected to SMS... now it’s just “great everything is getting more connected and homogenous and I seem to have less and less options”
I think with a smaller product you're trying to fight for market share and retaining a client - Microsoft is so large and with so much momentum and dug so deeply into their customers that this sort of thing doesn't cause them to want to find something else. If it hurts to change, it's easy to excuse something like this.
I think it's more "momentum". With a smaller company that affects a small number of users, the possibility of cutting your losses is quite feasible. On the other hand, with Outlook you've got your whole org using it, your servers running on it, your customers integrated with it, custom rules setup - a cut and run is just not simple.
Honestly though, I really don't get why rollback isn't more common for these larger pieces of software. Just some logic to detect multiple crashes, check if a rollback would break anything (i.e. dependencies) and then rollback. If you remove so much power from the average user in your OS, you ought to really have some amazing automated procedure for these scenarios.
I thought they also do limited roll-out as well? I think I remember the Windows anniversary bricked a bunch of machines in the limited roll-out and they just pushed it out anyway?
The complete lack of control as to when an update is performed is astounding - I know there are some settings to stop this, but apparently there are cases where that setting reverts back to auto or you need to switch it back on in order to manually update. I remember the last straw for MS and myself was accidentally clicking update just before a meeting whilst on very low battery, with no way to back out and no charger on hand. It didn't end well.
These days if something is "Windows only", it goes in a VM along with other software I don't trust.
I don't really know of any small services I gave up on because of outages more or less than I would for any other service regardless of company size....
The complexity of a product probably makes folks less likely to 'walk away' but I'm not convinced they do more or less for any other given company based on size alone.
MS definitely has customers like hospitals and the military that will be very upset and frantic in any outage.
At least in the days of yore MS distinguished itself with support for that kind of client, famously sending service technicians in helicopters out to remote installations.
It would be interesting to hear, in the new era that they're in, if that kind of thing continues.
One angle to look at this from is that if Slack or GitHub goes down, there's a good chance your competitors are being held up too. It's like an armistice on (your corner of) capitalism. Whereas if it's just your system, you've been put at a slight disadvantage.
Also, unless it's happening every week there's probably an element of, "well it doesn't make me regret the overall choice to use this service, so there's no sense being upset about it".
The other opposite angle to look at it from is that if slack or github goes down but you aren't using them then you have a slight advantage over your competitors.
Most of the customers are too dumb to really care, and if they did, what would they do
It will take you 3 years to unwind your Microsoft business. Few people with that level of rage over poor Windows/Office software can maintain it that long. Normally, companies yell at the IT Staff, unless it's a really bad problem, then they yell at the TAM or account manager, who takes some abuse, wears an appropriate sad/chastened face, and talks about the Microsoft release ring framework, and that most problems are your fault because you don't follow a similar model.
In my experience recently with O365, these issues usually are an update issue related to some regression in Windows 10 combined with an Office issue, or changes to Microsoft's authentication infrastructure. Microsoft has a unique understanding and methodology for making things good enough to persist -- they don't need to sell the product anymore, just make it not suck too much.
The first issue is because no human can keep a product working that has like 50 different active releases while the company prioritizes pushing out slop. The second issue is usually tied to a product release or revision, combined with new infrastructure. (Especially if it's in new IP space)
If a service has a big enough base an outage could almost seem like a pass for everyone. “Oo couldn’t do that this morning GITHUB was down.” Whereas if my small GitLab server was down bosses could be “well why aren’t we on GitHub like everyone else, it’s up now?”
When google calendar was down a bit ago I tried looking at random people responses on twitter and many were “whew I get out of some meetings!”
Perhaps there is some accountability for large scale outages like these but it really feels like consensus is often “shit happens” which is totally reasonable just seems like it could hurt big companies a lot less.