> the very fact that The Pentagon thinks that all its cloud apps are going to work under either AWS or Azure shows how ignorant they are
I'm not sure why that is the case. I worked at a place that mandated minimum two cloud support and we were going down that road when I left. I didn't see any complete show-stoppers from a technical perspective although there were a few annoying issues. Maybe the author is just hammering home the incompetence angle, where the IT managers he lampoons are incapable of managing such a project. But at it's face, holding an expectation that systems be redundant across cloud providers seems reasonable.
I didn't read it as cloud applications should be interoperable between providers but that a given application should be able to pick which platform they build on. I've had a similar argument with a previous employer who wanted to select just one provider. I was baffled at the time as provisioning an account on any of them takes seconds. So there's no reason to pick one other than politi---and right about then is when I understood the whole motivation.
As one of the "Federal IT managers" actually doing this, there are too many broad statements here that the overall message ends up being misleading.
If you were completely cynical, then you could find enough examples to make anything in here seem true.
For example, this entire paragraph is wrong:
Here’s something that will surprise you a lot: when it comes to government, cloud computing represents a huge shift of money from the public sector to the private sector. It’s the privatization of of government data. Lock-in is completely ignored: how will government departments ever get their data back out of the cloud? “Not my problem,” says the federal IT manager, “besides, there’s nothing about lock-in in these Powerpoint slides.”
There is realistically too much here to unpack in a comment, but I would say that the overall thesis of the article is pointing in the right direction.
However it's not like failure is a forgone conclusion, if competent people (like a lot of you reading this are) join the government to actually help fix these things then we can actually do things correctly. I posted in the Who's Hiring Thread last month so we're ready whenever you are.
I've never actually looked into government tech jobs, but my assumption is that they would pay much less than a competent person could make in the private sector. Other than pay or patriotism, what would motivate a competent person to want to work for the government?
If you think that your government should be doing things a different way, and you feel you know how to do it better then you have a couple of avenues to participate. One of them is joining the government in a capacity that allows you to affect change directly.
It's true, you aren't going to get paid $500,000 a year for being a software engineer. However, the USG does not pay subsistence wages. For entry level Data Science jobs in Boston we pay better than the market, and with better benefits, a union, matching investment accounts, low cost health insurance, stability and many other things you don't have elsewhere.
We also have opportunities to work on things that you just can't elsewhere.
Let's be honest - working for the government is a very poor way to affect change in it. The decision makers are also very far removed compared to private corporations and certainly so compared to smaller more agile companies like startups.
I am sitting here, as a government employee, telling you that you can actually make change. And at least for this problem set, you can make it in an extremely impactful way.
In what way have you effected change in the government? In what ways have you seen those around you do so (you must have witnessed this, by your stated claims)? I was in the military, which is a different beast entirely, but worked constantly with fairly high level government employees. They always seemed powerless, and to have accepted that.
But in the federal government sometimes the decision makers are so absent that you become the decision maker. As in, the President appoints his golf buddy to run a big agency, that guy spends all his time with hookers and drugs and your agency is left to its own devices.
I'm sorry, I don't want people in the government unilaterally implementing change they happen to think is the right thing, anymore than I want a dictator.
If the government wants to attract competent people, maybe it should first get rid of mass NSA domestic surveillance, purge the CIA of everyone who was responsible for extraordinary rendition, and try harder not to murder innocent people overseas. As things currently stand, I feel morally disgusted just paying taxes in the US. I definitely would never work in anything that even somewhat came close to the military or intelligence agencies.
I guess this article counts as therapy, what with the whole line of "The goddamn vaunted databases of the government are NOT the stuff of conspiracy theories. In fact, they're just as shitty as you would expect."
Honestly this entire system is a mess. But it's working so nobody is going to change it, plus going "over the top" and building a better, more idealistic solution will have the same set of problems. Assuming you end up building something dramatically better or easier, getting market share means addressing more and more use cases, and you more or less end up as the n+1th protocol or standard.
To me the only long term in dealing with this mess is going to be some shift in how we actually do computing on data. Leave the data at rest / in-situ and move more and more compute capacity to where it sits, then merge results together later. We're getting to the point where containers are common place, and FaaS is becoming comfortable.
Why do I have to swipe an access card and display a name badge in my building, when all the important data is outside the building?
Why do we factor GDPR into our designs if we don't know where we store the data, and we'll never meet (nor be able to trust) the people who hold onto it for us. Can't we just encrypt it on our side then? I don't think we're getting homomorphically-encrypted relational databases anytime soon.
My last month has been an effort of 'migrating' a service from another team to ours. The service stayed right where it was - the cloud - but we sank weeks into editing IAM files and deploy scrips to try to make it 'belong' to our team.
We're programmers; we don't know about AWS's policies model, security groups or software-defined networking. Whenever I'm forced to interact with AWS I always feel like I'm doing significantly more work than the "managed" selling point of AWS would imply.
I know my way around ssh, docker, iptables etc. But I miss having someone in the team whose actual job it is to be good at these things.
The best part is the elasticity of cloud storage. When the projects fail, they'll just keep all the data in the failed project achive. The next go will have it's own multiple copies of the same data and so on. They'll just keep paying incremental storage charges. Meanwhile, behind the scenes in the cloud - automatic, transparent deduplication....
Sure: "The cloud doesn't solve common IT problems, only shifts them around, and makes some problems worse, such as more vendor-dependency. If you hire amateurs, you get amateurish results. Renting cloud-based amateurs has all the same problems as in-house (internal) amateurs."
They won’t admit it, but many companies move to the cloud not only because they don’t want to deal with administering servers but also because they don’t want to deal with server administrators.
It’s not like on prem server administrators have a great track record when it comes to security.
I'm not sure why that is the case. I worked at a place that mandated minimum two cloud support and we were going down that road when I left. I didn't see any complete show-stoppers from a technical perspective although there were a few annoying issues. Maybe the author is just hammering home the incompetence angle, where the IT managers he lampoons are incapable of managing such a project. But at it's face, holding an expectation that systems be redundant across cloud providers seems reasonable.