They don't do it because it may or may not be right for their needs. They pick Azure because somewhere along the line the mid- and upper-level managers, including director-level IT people, have been made to believe that anything other than Windows is insecure and not suitable for HIPAA.
If I even mention the word "Linux" or "AWS" to them they immediately curl up in a little ball mentally and treat me like I'm some kind of evil script kiddie, or magical wizard. It's truly baffling the responses I get from people who are supposed to be in charge of these things.
I don't know these people outside of boardroom presentations, so I don't know how they got this way. My best guess is that it's either Microsoft salesmen, industry conferences, or simply what they're comfortable with and anything else is not worth risking their jobs.
Lastly, all three of AWS, Azure and GCS do try and make a big deal of their HIPAA compatible instances, but whatever the compliance officer at a company is used to is what will get by the easiest - honestly anyone who has a compliance officer that is happy to accept a non-on-premises solution should count themselves as lucky.
All that said - I'd be amazed if MSFT hasn't invested heavily into advertising to cement this decision, but I'd also be surprised if AWS hasn't been trying to dump even more money in there - google just seems a bit clueless when it comes to the convention game and these sort of closed circles where everyone plays golf on tuesdays.
I'd argue it's less because of a belief that a non-Windows environment is inherently insecure and more that Microsoft very aggressively promotes their products as being compliant with regulations like HIPAA.
This isn't limited to PaaS. For example, I think that Teams will likely gain a foothold in healthcare environments over Slack because MS markets Teams as explicitly HIPAA compliant. (Slack explicitly is not).
Things like that buy them a lot of mindshare when it comes time to choose a PaaS provider.
Oh, and Slack is available with HIPAA compliance — https://slack.com/help/articles/360020685594-slack-and-hipaa
Also, as someone who actively shops for SaaS offerings for healthcare users, I can say that AWS and Azure are some of the very, very few organizations who offer anything with HIPAA compliance that isn’t locked away at the most expensive “contact us for pricing” tier. It’s so annoying. Never a chance to try a service with a low cost/stakes pilot/prototype because they hit you with the full sales pitch and highest price points. Being able to dip our proverbial toe into AWS with a low time/dollar commitment was a huge win.
What Microsoft sells is a "good enough" solution that speaks the right language for the decision makers. In turn, this makes their decision less risky while still satisficing the requirements.
If you have an existing business relationship with a company and already trust them to help you navigate regulatory hoops, there's going to be strong inclination to continue to trust them as you expand your infrastructure to the cloud.
Like you said, it's the kind of very smart, long-term marketing I tend to associate with "New"/Nadella Microsoft.
I still don't trust them, but credit where credit is due.
Namely, if you were to state to a higher level manager that you "don't trust Microsoft", you will come across as a crank. Or a developer who doesn't see the bigger picture.
To improve perception, folks should try to reframe their personal preferences ("I don't trust MS") as risks to the business - "I worry that MS will lock us in to their platform and jack up prices". The former is reputation-harming; the latter is being a savvy 2nd order thinker, who looks beyond just what the MS salesman says.
The next one would be the PRISM connection (Skype architecture, etc), and FISA orders they're surely under.
It's not totally they're fault I don't trust them. Buuuuut, it is partly.
Definitely appreciate the feedback - lucky for me, I have removed myself from corporate tech bureaucracy, become self-employed, and can preach from my poor man's pulpit ;D
As a result what we end up having is this shadow-structure/architecture at the company that is highly resistant to change, such that when new projects are discussed and designed, oftentimes decisions are made that are technically worse but save time due to not messing with the SOX structure.
I feel the Azure choice might be due to a similar phenomenon. Maybe the Azure people promise all sorts of out-of-the box HIPAA compliance and insurance and security etc. and the upper management is happy not having to deal with it. Price difference almost doesn't matter very much, if saving a few bucks means that you'll now have to hire and retain a team to design and maintain compliance in your AWS stack. I've no idea if this is actually what happens but it smells like it.
SOX is the standard for change management for financial companies to prevent shadow changes that could enable another Enron, Worldcom, or Adelphia style cooking-of-books financial scandal. The idea is that all changes should be fully analyzed for their impact. If a change is found to have been made without going through the controls process then a review and an emergency after-the-fact change record must be created. This leads me to believe you did not just login to a server, unless you weren't supposed to have access in the first place (separation of duties requirement), but actually made a modification to that server without an approved change.
All this means that the system is very resilient and will not suffer random outages but most importantly, in the SOX world, there is no avenue for malicious changes to be introduced to critical financial systems and their data.
Even good and thoughtful regulations get interpreted into nonsensical bureaucratic nightmares by some companies’ compliance departments. The implementations that add disproportionate friction usually don’t achieve the intended benefit either. It’s not people choosing a convenience vs. safety tradeoff. It’s just people who don’t understand the subject matter at all.
I work with hundreds of vendors and digital health organizations that deliver life critical software applications to health systems - and we host everything on AWS. Independent 3rd party HIPAA assessments + a HITRUST (not self assessed) certification goes a long way with hospitals now.
Also, Microsoft support for enterprise customers is just amazing. They have been deploying their own engineers to support enterprise customers in the past for their hosted clouds.
Cant mention much about security but Microsoft does open fewer lower level Apis compared to Linux. For example, it's impossible to do certain socket-level ops with Windows apis. So, in a sense I guess it makes Windows safer at least historically?
If lacking a ribbon interface makes you feel like someone chopped off your hands...take your hands off the mouse and put them on the keyboard, haha
People are professionals, and these decisions are important and expensive enough that they are very rarely made by lower level management. Sure your CTO will advice the executives to go with Azure, but it’s not because he or she fears Linux, Azure runs a lot of Linux after all. It’s because of several business related reasons.
When you’re and enterprise sized organisation your IT staff probably knows Microsoft products. They are certified in them, and have worked on them for maybe decades. You probably also already run an Office365 platform, which means you probably already run things like Azure AD, which again means, that training and re-training costs will be millions lower if you stick with Microsoft rather than going with AWS, and Google is typically not even on the table because their support isn’t competitive with Microsoft or Amazon. Not only are there the training cost, but you will probably lose your best IT staff members if you don’t pick the technologies they like, and that’s really expensive in a world where some of the hardest IT people to hire are people who know how to manage an enterprise sized cloud setup.
Then there is the business relationship. We’ve worked with Microsoft for what? Three decades? And while the average Joe was making M$ jokes, they were the only company, aside from IBM, to take enterprise organisations with a lot of different business related needs serious. It’s hard to compete with that sort of history. Amazon actually does a great job with AWS now that they’ve realised just how much money is available in public sector cloud, but they’re still up against a business relationship where we know that we can call directly to Seattle when something breaks and that they will be on the phone with us all the way until it’s fixed. Not having the same relationship with Amazon is a huge risk factor in your strategic analysis, and that is something most enterprise organisations but especially the public sector is very careful with.
I don’t think I’ve ever heard a single argument to run Windows because it was safer than Linux, or any of the million other Internet armchair options on why the public sector wastes its money on Microsoft. The truth is a lot simpler, there is no alternative, and Microsoft is actually very good at selling software solutions as well as support to enterprise organisations. There is no actual alternative to office365, and 90% of the 300-500 applications/IT-systems you operate only run on Windows and have no alternatives available. Like we run a medical journal that keeps track of first graders vaccination history, it’s client only works on Windows and it’s backend is on a mainframe. We’d like to replace it, because only a few companies know how to operate mainframes, which means they can squeeze us on price, but no one has made a competing system, and that’s just one of a million such stories.
If the price difference isn’t significant, you’re going to have a very hard time competing with Microsoft if the enterprise organisation you’re selling to in anyway has to operate part of its backend in a lot of organisations.
All that said, cross-cloud data links are relatively easy to establish - just like linking in data from someone else when everyone had on-premises servers.
> That report notes that "insufficient interest" from Epic customers in Google is behind the decision to focus efforts instead with those cloud competitors.
It seems like if you run Epic Systems at Google, you were in a minority that is apparently not worth supporting. This seems less like Epic putting their thumb on the scale, and more like their customers putting their thumb on the scale.