Neither of them should get it.
Government services should be provided from government data centers operated by government employees.
Private companies should NOT be involved at all.
The US Court for Federal Claims on Wednesday denied motions by the Department of Justice and Microsoft to dismiss an Amazon Web Services lawsuit challenging the Defense Department's decision to award its Joint Enterprise Defense Infrastructure (JEDI) cloud computing contract to the Windows giant. Since 2019, AWS has been …
Should the government run build its own tanks? Guns? Ships? Aircraft? Heck, if we believe that the government uses AES, then it doesn't always roll it's own crypto. And DARPAnet was contracted out in the first place.
The issue here is that the government pay scale does not allow them to pay enough to attract the talent to build the system that they need. There is just no way for the US government to do this in house.
Amazon has been crying in their beer because they tried to buy the contract at the wrong level, and u$ showed them up for being the newbs at government contracting that they are. Government contracts, so far as I can tell, have been a mess in this country at least since the first congress ordered ships of the line. (And I have no expectations that the Confederation was any better, I just don't have an example.)
> [ ... ] how do they think that a subcontractor can do if for LESS [ ... ]
They don't think that. It's a form of corporate socialism.
At a minimum, we could save hundreds of millions of taxpayer money bypassing defense contractors' CEO's and upper-level suits' salaries and bonuses, while keeping everything else equal.
Actually it is not the pay scale but the what the IT bods do that is important. Where I am I do custom internal programming in an obscure language. Having our IT department build a cloud system would be foolish because none of us have the background or skills to do it efficiently or at a reasonable cost. Our day to day activities are doing other things that need to be done so hiring/using an outside cloud vendor has serious merits. The government is in the same situation, their primary IT needs do not revolve around many activities thus they have very limited experience or skill in many areas.
"Amazon has been crying in their beer because they tried to buy the contract at the wrong level"
I'm not sure it is just Amazon that are crying here - current rumours are that JEDI will be superseded by a new tender as there is no way that the current JEDI contract can deliver savings given the blow out in cloud service usage over the last 2 years and that awarding the contract to Microsoft has meant that the DoD now has a huge dependency on one supplier as O365/other contracts were awarded to them already.
So who does win?
- IBM/Oracle/other legacy data centre operators won by getting another 3-5 years life out of their DoD contracts - Oracle in particular were sitting on a significant cloud investment that could have become a large loss. While they may have been able to wind down their investments, the DoD will still likely reduce the number of vendors so its still a partial loss.
- while AWS have currently lost the contract, they are likely being paid more for legacy services (and growth in those services) than they would have got under the new consolidated contract. It was believed that JEDI would flatten DoD revenue for 2-3 years at around $600m per annum.
- Microsoft may be doing the worst out of JEDI as it currently stands as they have made the $200m-$300m investment in new resilient data centres without getting the full benefit of the $10bn over 10 years of business. Yet. Or possible never. But they have existing government O365 in resilient facilities (something they had to deliver for DoD eventually) AND they will be on an even footing with AWS in future tenders versus being around 3 years behind AWS in terms of DoD-compliant facilities so it's not likely they do badly, they still get their cake, jst some of the icing is missing.
- DoD IT have lost out on attempting to keep their budget at around US$20bn per year for 2020-2025 and addressing legacy security issues in older facilities as quickly as they would have liked. In the longer term, it has likely lost the JEDI battle but won the war against some of the legacy vendors that have caused the huge increases in DoD IT budgets.
- US taxpayers. Congratulations! How much have you won? An additional $1-2bn per annum in DoD funding? Or is it more? Future numbers from the DoD will be revealing but with the push to cloud already happening and COVID likely delaying the decommissioning of legacy facilities, It wouldn't surprise me if the US$200-220bn the DoD expected to spend between 2020-2029 is now closer to US$300bn. But just think of all those DoD suppliers whose kids you have paid to send through college, who can now afford that third luxury car in their driveway and a winter AND summer holiday home. They're taxpayers too right? And maybe some of their wealth will trickle down...
- Lawyers. With the JEDI contract now ruined, what could have been a prosperous next 10 years will likely be curtailed as it's not in any ones interests to fight this. The AWS/DoD/Microsoft service love-in will continue and they will think of a better way to get the result they want - AWS/Microsoft were small players in the DoD IT budget and are replacing the legacy vendors who wouldn't play nice. Playing nice now means more of those DoD billions become yours and an actual fight between AWS/Microsoft can come later. When there's less alternatives.
"Government services should be provided from government data centers operated by government employees."
This was an argument lost 25 years ago when the DoD first started to privatise IT services - as JEDI is about both cost reduction and utilising the inherent redundancy available from the very large cloud providers, I doubt this will change now.
JEDI is for a completely new set of datacenters. It doesn't matter who does it--Google, Amazon, or u$, they will NOT be using GCP, AWS, or Azure. Certainly, the providers will bring technology. Certainly, lessons learned will go back & forth. But at the physical layer, JEDI was/is for an entirely new set of datacenters.
Even the most trivial of security concerns demand it. Or the most basic of DOD-level survivability concerns.
This post has been deleted by its author
The trick to pork barrel contracts is to bid low, get the gig, and then when you've completely queered the pitch for anyone else suddenly double, treble, quadruple your fee.
So, if you want to make (say) £10 billion from a contract, you bid £3 billion. Get the gig (and politicians will love you as you allow them to appear as if they are saving taxpayers money). Run it for a couple of years and then turn around and go "oh dear, unless you pay us another £7billion, this project will fail".
Rinse and repeat.
I invite commentards to reply with UK government contracts in this vein, starting with Universal Credit.
You've got a point, for consultancies the magic phrase is "change control".
The process is so transparently bent I'm still astonished they keep getting away with it, even now.
It goes in phrases:
1 - a government contact writes the RFQ. Because they typically lack the expertise for it, the consultancy offers "help" to write an "industry standard RFQ".
2 - the bidding starts. The consultancy now has an insider, but pretends that the whoever in their setup is bidding is "independent" and the people never meet in the pub and over dinner to align their stories, no gov, that's just a vicious rumour and look, we even have an ethics policy.
3 - the consultancy wins the bid by going low. Now comes the hard work in maximising their take. It starts with "reserving resources", which translates in paying people to be on standby (yes, not to work yet), then growing the team to the maximum they can get away with, even if it means some people are mainly making tea because there's no plan yet, but hey, they're "industry leaders" so they need to be paid well. I've seen people so fed up with that that they asked to be allowed to leave the project, which tends to be denied.
4 - hello change control. This is where the "help" in point (1) was crucial: by some amazing coincidence, the really critical parts are ill defined and need to be improved/changed/whatever term applicable that results in more billing. And hello project overrun that was totally unpredictable and circumstances have changed and it all needs to be up to the highest industry standards (and, and, and - the BOFH excuses list is but a patch on the BS they can spout to get their hands on more money).
5 - frequently but not always, the project is eventually abandoned as "previous administration", "no longer relevant" and other terms that allow the utter waste of money and absence of any actual deliverable to be buried six feet deep, never to be seen again.
Rinse and repeat.
Also, it helps if you have a previous senior member of management of his outfit run the very agency that has to audit the project. I've seen that.
Everything you say is correct - it's the game IBM/Oracle tried and lost at with JEDI.
IBM/Oracle tendered for the contracts with the proviso that the contracts would start once IBM/Oracle had purchased land (with government help naturally - both paying for them and taking the property from current owners) and the DoD had paid for new facilities to be built (minimum of 3 separate facilities in two separate states). Which the DoD would have paid for separately to JEDI. And delivery of JEDI services would have started when these were completed. In 3-5 years. When their legacy DoD contracts would have been rolled over for a few extra years at additional cost and no additional investment....
Microsoft had 2 government compliant facilities in one state when the contract was tendered but has since built out the other facilities at their own cost (and I believe they completed these in late 2020).
AWS already exceeded the requirements as they had been providing JEDI-type services since 2015.
The really amusing thing is that the requirements for JEDI-type services were written by legacy DoD IT vendors to keep AWS/Google out of the market back around 2010 AND resulted in a huge effort to address "all of the weaknesses" in the legacy vendors security practices. By address, I mean highlight and result in legacy vendors asking for billions to address their non-compliance.
So, we want our national compute jobs not going to a software company but to a logistics company famous for maximizing profits and the occasional self-own.
JustTaxes Blog: "Amazon has Record-Breaking Profits in 2020, Avoids $2.3 Billion in Federal Income Taxes." paying jus 9.4% in taxes and not 21%.
So, Bezos started Blue Origin and blows a lot of big things up.
Weird flex for a logistics company but ok.
Why are we trying to send humans to Mars again?
For the writeoff?
"It would be great if the DoD could counter-sue Amazon if Amazon loses and reclaim litigation and court costs."
Hint: they only way the DoD has to save money for JEDI is if they use AWS.
While AWS maybe the bad guys for most things, this is not one of them.
This was the expected effect of JEDI on cloud spending in a 2019 estimate where JEDI had started:
Naturally, JEDI hasn't started so the 2020/2021 figures are likely to be north of $2bn.
For DoD facilities, Microsoft were ~5 years behind AWS and hence most existing cloud migrations from legacy facilities over the last 5 years went into AWS. Work that doesn't benefit from JEDI pricing and needs to be re-done if Microsoft is chosen. In my not so humble opinion, the AWS work would likely not be done as it is significantly cheaper than the legacy facilities that the DoD is using.
I would also suggest that Microsoft may struggle to match DoD demands - they have entered the Virginia data centre space race 10 years too late and latency kills performance...
Biting the hand that feeds IT © 1998–2021