What would you suggest as the alternative? Have DISA write their own tech stack, operating systems, and cloud services? This is a freaking great deal for the government.
I don't have any real numbers to back it up, but I have been told that Red Hat costs the government more money per machine for the licenses. At least within the DoD. Now this could well be attributed to buying less licenses, and not getting as much bulk discount. But Red Hat not necessarily a cheaper option. And even if you got the license costs down you would still have to pay absurd amounts in retraining the entire workforce.
DISA is a big group. They have mountains of RedHat instances already as well. I would bet they've probably got a billion dollars or so invested in the RedHat infrastructure.
If I had to hazard a guess without seeing the SOW/PWS, I would bet that most of this MS contract would cover desktop support services.
DISA supports IT services for a huge portion of the DoD. This is basically a Fortune 100 company in terms of employees and scale.
Are we talking DoD or DISA? DISA is probably close to a Fortune 100 company. DoD's annual spend exceeds Apple's market cap. The annual research budget is $40B+
Holding them ransom? They chose this at the highest levels. I was in DoD for the transition from Sun to MS on the desktop. You're off by orders of magnitude. My one, tiny little facility had an IT department of several hundred to support the crazy infrastructure stuff we had to deal with.
And don't assume that they're not also using other services and providers as well. How would you provide desktop services and solutions to 80,000 people distributed all over the world?
If you have a solution that doesn't cost $1bn, please write the white paper. I'll gladly write the RFP response and we can enjoy our profits. :D
The numbers may seem staggering, but this is how much shit costs in the government. Most of our competitors in my space are offering services with 2%-4% margins because the government is so stingy about how the work is performed. DISA alone is probably at nearly the scale of AWS in terms of compute power and system requirements. And they probably have 5-10x the number of employees. It's not a trivial problem to solve, and I've been working for contractors for 10+ years now.
I'd start with a base image of an enterprise Linux on certified configurations of hardware. This could be a VM, CD they'd burn, or Live USB. Specialize it to some more for specific roles. Semi-automate that part esp for creation, testing (esp updates), and distribution. Host it on those cheap, storage sites with verification & other tools pre-installed on their desktops by local admins.
One UNIX admin at each remote site with standardized tools for the various functions like DNS, email, etc. A centralized solution for communications, workflow, secure sharing, and so on. They just access it over the internet with replication to a few sites (or even all filtered by whose present there) to get availability or extra performance.
I'm not seeing this cost $1 billion even if I did it with VMS clusters. Even that overpriced system that ran whole enterprises was under $100,000/yr licensing per branch with mega-fee for main offices. I can't guess how much support would take but imagine local admins handling a portion of it. If it's just Linux desktops, extra load might be offloaded to vetted consultants or companies that do it on the cheap with anything they do logged. Admins or local developers come up with design improvements or tools for recurring problems.
>I'd start with a base image of an enterprise Linux on certified configurations of hardware.
And right there you lose, because I all but guarantee there is Microsoft-specific code running at all levels of this organization. Everyone has to be retrained, a lot of infrastructure has to be rebuilt. All of that will chew up your 1bn quickly.
This isn't some SV startup, it's the government. Typical "move fast and break things" mindset doesn't fly here. You don't get any "tear it down and start over" moments, you get to make those hairy migrations to ensure everything stays up with smooth, staged cutovers. And that's not even mentioning the red tape...
I don't do Silicon Valley philosophy. I'm talking the efficient stuff that midsized organizations overstretched IT teams have been pulling off for years plus some tooling that comes with DISA's extra money. However, the lock-in effect...
"because I all but guarantee there is Microsoft-specific code running at all levels of this organization. Everyone has to be retrained, a lot of infrastructure has to be rebuilt. All of that will chew up your 1bn quickly."
...is a valid reason to default on using Microsoft. The difference is that moving what I can to alternatives means I can gradually move from lock-in to more flexible IT with open standards & multiple vendors to choose from. Organizations on Microsoft are often stuck, though. This much is undeniable. IBM, Oracle, and SAP the other huge offenders here.
It was every time I did it. Plus many of the individual tools are free and mature. No licensing fees. There's also the benefit of not having BSA looking into your firm.
The best benefits, though, are in flexibility and security rather than cost. I just have so much more available over long term to me with open formats, protocols, API's, and code. Plus, Microsoft has been saying "screw you" to its own customers in product development over past few years leveraging its lockin to keep them. Whereas, open stuff gives opportunity of switching vendors.
