Axed $170 million contract shows how DOGE-led cuts came over US Navy objections

[Yo_Mama_Been_Loggin]

A US Navy contract to move sailors' server-stored records to a secure cloud system was recently torpedoed as part of DOGE-led cuts that show how top officials are under pressure to find large cost-savings even over the objections of their own organizations.

An IT services provider named Pantheon received a $170 million contract last year to relocate the records threatened by flooding from a Tennessee data center to cloud storage. But a top Navy official ordered it to be cut, at the suggestion of Department of Government Efficiency newcomers, over the strenuous warnings of their own personnel officials.

An internal memo reviewed by BI highlighted that the system that DOGE recommended reverting to has been plagued by delays, a bloated budget, and little to show for it all.

Continued "delays have resulted in the Navy having to expend even more resources on legacy systems that are past end of life and do not meet the needs of the Service," the memo said.

https://www.msn.com/en-us/news/us/axed-170-million-contract-shows-how-doge-led-cuts-came-over-us-navy-objections/ar-AA1FS4Fj

Might be penny wise, pound foolish.

[lapfog_1]

1. this happens to have been my specific area of expertise for decades now

starting when I funded ( via my time at NASA ) the Berkeley RAID project in 1989.

So lets do some fun with math...

There are currently ( according to google ) around 600,000 active duty, reserve, and civilian contractors in the US Navy.

Assume ( for the sake of easy math ) that every last one of them either changes or accumulates more data every year... and we want to store at least 60 years of data. Let's say the amount of data per person is 1GB/year. How much data are we talking about.

600,000 x 60 x 1GB = 36 million GBs...

sounds like a lot, right? Not really... 36 million GB is about 36 thousand Terabytes... which is 36 Petabytes.

Normally you move things to the cloud to take advantage of the erasure coding applied across geographically distinct regions... let's say a decent 4+2 erasure coding... or 6 regional data centers each holding 9 PB. 9 x 4 equals 36 but we need 2 redundant erasure coded segments equaling the 9 PB at each of the six. We can lose any 2 data centers out of the six and still be OK. Just for kicks we will also buy long term archival storage from someone like Iron Mountain ( usually tape and stored in climate controlled vaults underground ) for the entire 36 PB. So... all told we need 6 x 9 PB of active storage or 54 PB spread around to 6 data centers plus another ( cheaper ) 36 PB for what we put into our underground long term. We also need to account for the fact that we will grow at a rate of 600 TB every year and that we need to copy all of the data to new technology every 5 years ( both for allowing for online and deep archive storage ). That's 90 PB plus .6 PB a year for 5 years or 93 + 1.2 for the redundancy factor or 94.2 PB AND we need to double it for the once every 5 year copy so.. call it 200PB total

Using Amazon pricing for HDD storage ( online ) even though this absurd amount of storage is not all needed at once ( 1/2 is only needed once every 5 years for 6 months for the copy forward ) AND 1/3 of that 1/2 is in a one time offline archive site at someplace like Iron Mountain.

Amazon FSx offers HDD storage at $0.013 per GB-month and you get $312 M dollars over the entire 10 year pricing.

OK, lets do the more realistic math.

200PB is really only 100PB with a excess charge of "1 year" for the copy ( I know, I said 6 months ). so 1/2 of this total plus 1 extra year... so 156 Million plus another $15M for a total of $171M over 10 years... Now remember that 1/3rd of the cost cost to the long term archival copy... so cut $57M off but add $5.7 million ( long term archival on tape - the suspenders AND belt safety ) and you are at a cost of $171M - 57 Million or 114 million plus 119.7 Million. Over 10 years...

Again remember this for 1 GB of active duty, reserve, plus civilian contractors PER YEAR. Home movies, photos of every paycheck, health visits, accidents, promotions, duty rosters every day, etc. A GB per person is a lot of data, not including any de-duplication or compression.

Now we need to add in some people to transfer the data... some will be computerized on ancient systems, some might even be on paper. Lot of records to process... maybe 200 clerks for a year or more to capture the initial history. That's $24M for at least a year plus 1/10th of that for the remaining 9 years ( not counting inflation ) or a total of $45.6 million for people.

Total to do the project - $165.6 Million over 10 year life cycle.

But, since I do this for a living. I could do it for almost half ( don't pay Amazon ). Build and operate your own data systems at 6 locations around the USA... Small ones, like 1 to 2 computer racks at current density... plus the long term archival at Iron Mountain ( lets not re-invent something as cheap as they are ). I would bid the project at $100 M for 10 years of initial capture and operations, including significant growth over the 10 year life cycle and at least 1 technology change ( copy all data forward ).

Not cheap, but not what the contractor bid.

BTW, in case anyone thinks I can't do this... I already did! I ( and my small team ) at NASA replaced the prime contractor ( $500 Million proposed cost ) to create the climate data repository for NASA in the late 1990s ( EOSDIS ). For a fraction of the cost. At the specific direction of Vice President Al Gore. The original and successful "make government more efficient" person ( but something he was never given credit for doing ).

I've been doing large scale data archives ever since. not just large scale but high performance.

Still, all in all, the project should not have been cut.