4 ms·
In spite of appearances, I'm not sure if they went with existing USPS web resources, at least on the frontend. Looks like custom build using a very typical mode
by paulsmith 5y ago
In spite of appearances, I'm not sure if they went with existing USPS web resources, at least on the frontend. Looks like custom build using a very typical modern managed AWS services stack. In any case, interesting new state of the art for government, compared, let's say, to the HealthCare.gov launch.
At first I thought it would just be a new SKU in the USPS.com website. But it looks like they rolled something custom, using CloudFront, S3, Route53 (two new domain names, special.usps.com and special-api.usps.com for POSTing the order form JSON to), API Gateway (and probably Lambda and DynamoDB):
https://docs.google.com/document/d/e/2PACX-1vQP9G1cWiICGOoBAjrUm6XGPK7ERwl2TrpxnWpJlkuAVdnsP2tkD9q77ZxFPvkDhtOby-cDjr9cfbPe/pub https://docs.google.com/document/d/e/2PACX-1vQP9G1cWiICGOoBA...
- danso 5y agoNot sure if USDS (U.S. Digital Services) team worked specifically on the covidtests.gov rollout, but USDS was largely borne out of the dev effort that rescued Healthcare.gov: https://www.theatlantic.com/technology/archive/2015/07/the-secret-startup-saved-healthcare-gov-the-worst-website-in-america/397784/ https://www.theatlantic.com/technology/archive/2015/07/the-s... > The spectacular failure of Healthcare.gov at launch led to the creation of what came to be known as the Tech Surge, a group of Silicon Valley developers who rescued the website from disorganized contractors and bureaucratic mismanagement. That group gave rise to the U.S. Digital Service and, to a lesser extent, 18F, two government agencies now working to improve the state of federal technology
- buscoquadnary 5y agoAs for 18F for all else wondering it stands for where there office is located on 18th and F street. That being said I think they should really really should maybe give some thought to renaming their group.
- llimllib 5y ago> Not sure if USDS (U.S. Digital Services) team worked specifically on the covidtests.gov rollout They did[1]: > The United States Digital Service, which grew out of the effort to rescue healthcare.gov, is also assisting with the launch. "We have the best tech teams across our administration working hard to make this a success, and we think we're well positioned to do that," [1]: https://www.protocol.com/bulletins/white-house-covid-test-website https://www.protocol.com/bulletins/white-house-covid-test-we...
- sybercecurity 5y agoThe US Digital Services was formed to prevent exactly that. The main lesson learned was "follow industry best practices instead of just paying a contractor who sounds like they could do it." Strange thing is this was supposed to go live tomorrow (19th). Though maybe they were going by Guam where it is already 1/19.
- llimllib 5y agoIt seems to be an intentional soft launch
- zild3d 5y ago> (and probably Lambda and DynamoDB) Curious why would we assume it's running on Lambda & DynamoDB vs any other compute & storage? Previous known gov projects?
- paulsmith 5y agoJust speculating, because it's using a suite of AWS managed services throughout (CloudFront, S3, Route53, API Gateway) and it's a common design to use Lambda functions with API Gateway. Knowing that this is a very high profile, high traffic site launch, and knowing something of the attitudes of the USDS team, they are likely to want to minimize the number of moving parts and services they have to manage themselves. Also, observing the interaction of the site, which all boils down to a simple HTTP POST of a small JSON blob to the API Gateway endpoint, I am (again just speculating) assuming they are wanting to do minimal processing and just stuff the JSON into a database as quickly as possible. It could be an RDS database, but that still seems like too much administration, and especially since DynamoDB can do highly scalable writes, it just seems like the least surprising choice. I could be wrong, and it would be great to hear the team share on record what's actually under the hood, but given what we can confirm externally by observation plus what we know about the mission and the team, I feel reasonably confident it is all managed services, at least on the frontend component we can interact with.