Hacker News Reader

Comment by mark_l_watson | Hacker News Reader

When I was young I worked in the defense industry. If I remember correctly, pushing back hard when not winning a contract was not done. I can’t imagine how much this would sour future contracts and future negotiations.

I admit a little bias since I preferred Microsoft winning this contract. I am a very happy Amazon and Google customer, but as a taxpayer I felt better about Microsoft winning this. This is just an opinion that admittedly is not based on any expertise in government procurement of cloud services.

View on HN

Yes. Similar experiences. Basically the position I ended up in was mandatorily selling out to Oracle with no other choice on the table. The key purchasing decision was whether Sun or HP got to bugger us for hardware. This was loudly shouted out to drown out the “why do we need oracle” question. This took months of noise and pretendering (pretend tendering) and all sorts of meetings where no one knew anything about anything that was being required. But everyone was doing Six Sigma of course do the cargo culting was justified by process. Eventually HP was pulled out of the hat because someone saw a nice shiny sales brochure.

In the end, £1m down the shitter, 9 web pages written in JSP backed with a full 42U rack HP/UX crate running oracle was seen as a success. The whole platform was scrap in three years. each http request made cost £83 in the end ($100ish).

And that’s defence spending in a nutshell.

No one says a thing to stop this happening.

> If I remember correctly, pushing back hard when not winning a contract was not done. I can’t imagine how much this would sour future contracts and future negotiations.

It definitely can and does happen for big contracts. It's what you do if you think someone rigged the process and want to make the point that you won't stand for that.

I heard several stories about exactly that when I was in the defense world.

> If I remember correctly, pushing back hard when not winning a contract was not done.

At least two high-profile companies (SpaceX being the better known one) have done just that very successfully in the last few years.

Whats interesting is the Amazon got delayed for like a year on their other government contract due to IBM contesting their win [1]. While it might have been different in the past, contesting contracts is the way of the world these days. And that was only $600M contract. A $10B contract is worth fighting over even if it costs them millions. And the government knew it would be contested no matter who won. MS would have contested it had they not won. Unfortunately, the legal fees are just another part of the cost of the contract.

[1] https://www.informationweek.com/cloud/infrastructure-as-a-se...

Isn't the problem that they are driving into a $10B contract instead of being agile and avoiding a single point of failure?

That is definitely part of the problem, they should have forced a multi cloud solution. Bring at least two vendors in, and make them compete and keep eachother honest. Either way, you need to keep the around for years, so you have time to get things set up to meet the security standards, migrate apps to the cloud(s), and all that. A 10 year contract does make sense as far as not having to keep redoing things. But as it is, that will end up a much longer contract due to vendor lock in.

I remember the Air Force Tanker deal, with the civilian official running the most obvious palm-greasing ever.

https://en.wikipedia.org/wiki/KC-X

Why is Microsoft better for taxpayers?

Because the investment prevents Amazon having a total monopoly on Cloud and helps keep the price from skyrocketing in the future.

I say that as someone who is all in on AWS.

This is an important point. It’s a readiness premium. Otherwise, you’re beholden to an org “too big to fail” like we’ve done to ourselves with Boeing.

I don't think winning a federal contract is what would allow or disallow Amazon from having a monopoly on cloud, and I don't think the current state of the market is monopoly (though oligopoly or duopoly might be appropriate).

Amazon already has GovCloud and C2S. If they also got JEDI, they would have complete lock in / monopoly for all of the federal government's cloud services. And with the length of all these contracts, they get guaranteed lock in for a long time. There needs to be competitors in the space to prevent complete lock in. Otherwise the price of re-doing all the systems for a different cloud would be immense. They could basically set whatever price they want on their services, as long as it cheaper than the cost of moving they will keep the contracts for decades to come.

MS and Google already have various government cloud deployments. Amazon just squatted on the "GovCloud" name.

And Congress can always pass a law to limit prices, and the Executive can easily win an antitrust case if the vendor gouges them.

> Amazon already has GovCloud and C2S. If they also got JEDI, they would have complete lock in / monopoly for all of the federal government's cloud services.

GovCloud isn't a contract, it's a region certified to meet FedRAMP-HIGH and other federal data requirements (and is available to non-federal-government customers that need that, as well as federal government.) Microsoft has an equivalent (Azure Government), and it's public regions are FedRAMP-HIGH certified. GCP also has FedRAMP-HIGH certification in many of their public regions but doesn't seem to have anything directly comparable to AWS GovCloud or Azure Government currently.

So, no, JEDI wouldn't give AWS a monopoly either on federal government cloud or cloud for people dealing with federal mandates like FedRAMP.

I did not mean to imply that GovCloud was a contract. I though it was the exclusive provider of isolated cloud services. I didn't realize Azure had its Government region. Glad that exists, I want to see that competition.

However FedRAMP is not relevant to what I was trying to say. My understanding is government itself directly uses just the AWS and now Azure isolated regions. What I was meaning to say was for the DoD, which I though was exclusively under AWS GovCloud, and AWS C2S, rather than the whole government. They could certainly be using Azure Government though.

I am glad to know there is more competition than I though. Thanks.

> My understanding is government itself directly uses just the AWS and now Azure isolated regions.

This is not true, at least based on the discussions I've been involved in with AWS reps (while my employer is a state government entity, how the feds use AWS has frequently come up.) There are US Government customers without the requirements that are only provided on the isolated regions that use the commercial offerings that have the appropriate guarantees for their data, and there are non-federal customers that use the isolated government regions for data which has the higher-level federal requirements that require those guarantees. FedRAMP isn't the only force here, but it's a key requirement that drives the choice. (Also, GovCloud was cleared for some requirements before commercial offerings, so some customers, including some federal ones, are currently on GovCloud with workloads that could be served on the commercial regions now.)

Microsoft has a pretty strong pitch for government cloud services. The department of defense runs one of the largest Active Directory deployments in the world, along with a huge install base of Windows desktops, Office, SQL Server and SharePoint.

Many of these DoD deployments have been extensively customized by Microsoft’s professional services division, which has an excellent reputation in the defense industry. Apparently Microsoft rolled lessons learned from these big deployments into their Azure offerings.

I think it’s clear that DoD can reduce costs and improve service quality by migrating to Azure managed service variants of existing Microsoft systems. Amazon is probably better for basic IaaS, but this SaaS/PaaS stuff is gonna be better on Azure, at least when you already have such a big commitment to the ecosystem.

There are probably some strategic concerns too. Amazon has been the only real cloud provider to the defense world for years. Giving Microsoft a big contract means the government will have more options in the future. This dynamic really only shows up in huge enterprises with weird needs... look at NASA for another example.

If you want to complain about Microsoft’s behavior, the best place to look is probably at the non-hosted upgrade paths for their legacy products. They could refuse to let you run Word without Office 365; Office 365 requires Azure, so you have to certify Azure for your data to keep writing documents. That’s a pretty aggressive move to pull on any client though.. the sort of thing that pushes folks to make do with Libre Office.

"MS runs a lot of propriety IT systems" isn't a reason I'd want government using their stuff.

Government should run on open standards.

Reply on news.ycombinator.com