I’m an enterprise technologist. An architect that’s been living in the enterprise for 17 years. I also happen to have a Healthcare background (specifically in IT systems and Health Information Exchanges). I have extensive payor systems experience (integrating insurance carriers, medicaid, medicare, Centers for Medicaid and Medicare Services (CMS)) experience.
I also happen to be interested in healthcare via the exchange. Here’s the bit that peeves me.
634 Million to build Healthcare.gov .. and it doesn’t work.
I’m going to take 15 minutes to dissect the budget for the project. Why? For the life of me, I can’t figure out where the breakdown is. Was the project under-funded? Uhm.. No way.
Let’s assume for a second that a management consulting firm was put in charge of this contract, it still doesn’t justify where / how / why the site is completely dysfunctional.
Let’s just throw some complete “pie in the sky” numbers at the topical dissection.
I’m going to be completely ridiculous with my estimates (as in the site was built by completely incompetent resources).
1st thing: Let’s just take 50% of the budget away and call it “profit taking”.
So this leaves us with ~$315 Million dollars. (I rounded down, because I wanted to be conservative).
Now the ridiculous numbers I’m going to throw at it:
$5 Million to develop the website. (A handful of pages.. oh hell let’s just throw $15 Million at it)…
We’re left with $300 Million dollars.
$10 Million to buy two buildings and outfit them as data centers. Not large ones, but not exactly small either)
$20 Million for Routers, Load Balancers, and Network Gear.
$10 Million for the labor to build out the Data Centers.
$50 Million in servers for 2 datacenters
$20 Million for 2 EMC SANS – Every possible option with a ridiculous amount of storage
$20 Million in Microsoft Licenses (because no government installation is complete without a few excel spreadsheets linked to a cluster of MS-SQL servers)
$20 Million in Oracle Licenses
$20 Million for a pair of Mainframes with all the trimmings, because no healthcare solution is complete without them.
$10 Million for Mainframe Software licenses
$10 Million in Internet Access (because bandwidth isn’t cheap right?)
$10 Million to interconnect to a clearinghouse and for the “integration fee”
$50 Million for “unforseen expenses”.. like Fiber cross-connects.
This leaves us with $50 Million dollars.
So my question is.. Why doesn’t it work?
Here’s the reality of what it really costs (and I’m being very generous):
$5 Million for site and infrastructure design and architecture
$5 Million for actual development
$10 Million for software licenses
$50 Million for Servers and a 2-3 cages in different locations
$20 Million for 4 SSD backed SANs
$10 Million for Deployment
$10 Million for Project Oversight
$10 Million for DevOps staff training, onboarding, testing
$10 Million for Clearinghouse integration and software licensing
$20 Million for 2 call centers with all the trimmings (and staff onboarding)
Even being completely ridiculous, I’m still looking at a significant savings.
Now, to do it right:
$5 Million for site and infrastructure design and architecture
$3 Million for Development
$3 Million for 2 Dedicated Cloud Environments
$1 Million to deploy the code
$3 Million to build 2 nice call centers fully staffed and trained employees
$10 Million just to have a little bit of wiggle room for unforeseen expenses.
Now, let’s take a few assumptions into account:
1. The application and infrastructure architects weren’t that good so we’ll just keep things up with load balancers.
2. The code is poorly written, so we’ll just have devops reboot servers constantly.
3. The Database servers are all on SQL Server so there’s going to be an “interface engine” in between somewhere, because as we all know.. RESTful interfaces are voodoo.
4. Integration into a clearinghouse or CMS to interconnect to the carriers is going to require lots of “batch” transaction processing (even with 5010 and ICD-10)
5. The ANSI X.12 Dataset was implemented poorly.
6. There are 2 mainframes somewhere in the mix running DB2.
Even with these knowns and unknowns, the $25 Million dollar estimate I estimated would provide an infrastructure that could easily handle 100 Million visitors a day.
As of the last estimate, there are only 313.9 Million people in the US.
The reality is, the whole project could have been done for less with more redundancy and availability.