It already uses a key/value store for on-disk storage, but you’ll have to write the server API and client yourself, along with a Raft state machine layer. It’s not a big lift though, and could make a fun weekend project.
It’s the first productionization of leaderless consensus protocols. It merges strict serializable transactions into the consensus layer (in 1RTT on the fast path), and has sharding built-in including cross-shard transactions.
It doesn’t support interactive transactions though, so it’s not a drop-in for SQL systems (but I don’t think they are suitable in modern systems anyway, given the latency cost). And it requires quorum reads, since it can’t use leader leases (for obvious reasons), so reads may end up being too slow for many georeplicated setups. But it’s a neat design.
Sanity.io | Backend/Golang Software Engineers | Oslo, Norway and San Francisco, CA | Full-time | VISA, ONSITE | https://sanity.io
Sanity.io is a startup building world-leading cloud-hosted content infrastructure at enterprise scale. Basically we provide a global, real time structured content database with powerful query capabilities and an open source framework to build highly customized content management interfaces on top of it.
We're looking for senior Go developers to help build our distributed document store. If you get a kick out of distributed systems, database theory, and query language design then we'd love to talk with you - prior experience with these fields is not necessary, as long as you're eager to learn. Position will primarily be based in Oslo, Norway, although we will consider applicants for our upcoming San Francisco office as well.
We offer competitive salary and stock options, a flexible work environment open to periods of remoting, a tight and friendly team with an awesome work/play ethic, base in the second happiest country in the world (Norway) in one of Europe's fastest growing cities (Oslo), and awesome social benefits as mandated by Norwegian law including five weeks of paid vacation, universal health care, child benefits, and one year of paid parental leave.
Sanity.io is a startup building world-leading cloud-based content infrastructure at enterprise scale. Basically we provide a global, real time structured content database with powerful query capabilities and an open source framework to build highly customized content management interfaces on top of it. We are now looking for talented backend and full-stack developers to join our fast-growing team.
We offer:
- Competitive salary, stock options
- Flexible work hours, open to periods of remoting
- A tight, friendly team with an awesome play- and work ethic
- Base in the second happiest country in the world (Norway), in one of Europe's fastest growing cities (Oslo) ticking all the boxes when it comes to arts, nightlife, and the great outdoors. (And everyone speaks English!)
- Awesome social benefits as mandated by Norwegian law including child benefits, paid parental leave, universal healthcare, four weeks of vacation
Sanity.io is a startup building world-leading cloud-based content infrastructure at enterprise scale. Basically we provide a global, real time structured content database with powerful query capabilities and an open source framework to build highly customized content management interfaces on top of it. We are now looking for talented backend and frontend developers to join our fast-growing team.
We offer:
- Competitive salary, stock options
- Flexible work hours, open to periods of remoting
- A tight, friendly team with an awesome play- and work ethic
- Base in the second happiest country in the world (Norway), in one of Europe's fastest growing cities (Oslo) ticking all the boxes when it comes to arts, nightlife, and the great outdoors. (And everyone speaks English!)
- Awesome social benefits as mandated by Norwegian law including child benefits, paid parental leave, universal healthcare, four weeks of vacation
Sanity.io is a startup building world-leading cloud-based content infrastructure at enterprise scale. Basically we provide a global, real time structured content database with powerful query capabilities and an open source framework to build highly customized content management interfaces on top of it.
We are looking for an experienced programmer to join the team developing our core document management layer. It supports deep queries in (optionally) schema-less data sets, fine-grained access control, real-time collaborative editing of rich data structures and scales to millions of documents. All our performance-critical subsystems are written in Go, with some line-of-business systems implemented in JavaScript/Node.
The work involves growing the capabilities and performance of our document query systems, real-time pipelines and maturing our ability to distribute these capabilities globally in a transparent manner.
We offer:
- Competitive salary, stock options
- Flexible work hours, open to periods of remoting
- A tight, friendly team with an awesome play- and work ethic
- Base in the second happiest country in the world (Norway), in one of Europe's fastest growing cities (Oslo) ticking all the boxes when it comes to arts, nightlife, and the great outdoors. (And everyone speaks English!)
- Awesome social benefits as mandated by Norwegian law including child benefits, paid parental leave, universal healthcare, four weeks of vacation
Much of the (valid) criticism in this article relates to the deuterium-tritium fuel cycle. This is the easiest reaction to accomplish on Earth, so most experimental reactors are designed with this fuel in mind, and we're certainly having a hard enough time making even this work.
However, I've always considered D-T fusion an intermediate step on the path to aneutronic fusion, such as Helium-3 or proton-Boron reactions. These avoid most of the radiation issues, as well as the tritium-breeding problem (although Helium-3 sourcing presents its own challenge). Since the fusion products are electrically charged the reactor could possibly also generate electricity directly, without a steam turbine and the associated energy loss. Unfortunately, it requires temperatures that are an order of magnitude higher than D-T (well beyond a billion degrees Kelvin), so we'll need to learn to walk before we can run.
The nice thing about fusion neutrons is you get to control the isotopes, you have no control over fission waste isotopes.
Some fission isotopes are really icky to deal with, as everyone has heard...
On the other hand if you don't like dealing with cobalt-60 waste at your fusion plant, simply stop using cobalt alloys in your reactor vessel.
It turns out to be "not that big of a deal" to design a fusion plant where neutron activation isn't important. The quotes are because nothing is easy in fusion but as a problem its pretty low on the list.
This is true. Back at Fiat Lux when we designed our D-D reactor, we intended it to sit inside a pool of water and borax. Since we didn't need to regenerate tritium, just absorbing the neutrons with boron was the cheapest solution. As far as I know, Borax is the cheapest effective neutron shielding known. We would have liked to have built our vacuum chamber out of purely Al (since Al-28 has a two-minute half-life), but we went with steel for cost reasons.
Unfortunately, we never made enough neutrons to activate anything worthwhile. Nevertheless, it is certainly possible to work around neutrons through design decisions.
Sure, activation in itself isn't necessarily a big problem, but I believe that embrittlement of the blanket and other plasma-facing surfaces due to the high neutron fluxes is one of the major engineering challenges for ITER and similar tokamak designs.
Billion Kelvin operating temperatures make me super skeptical about the practicality of that solution in my lifetime. It's just so many orders of magnitude beyond where our materials science is.
It's very high, but not as bad as it sounds. There is less than 1 gram of fuel in the reactor at any given time, so it's extremely diffuse, and mostly contained by magnetic fields. Much of the energy can also be radiated away by facilitating ion-electron recombination before the plasma reaches any materials.
The current design for ITER, operating at 150 million Kelvin, focuses the exhaust plasma onto a dedicated divertor surface with an estimated peak heat flux of 20 MW/m2. Experiments show that tungsten handles this fairly well, although with some cracks appearing after many cycles, but this is an active area of research.
At least with Google's GCR, this is no longer the case. They have a Docker credentials helper [1] which transparently handles authentication for regular interactive use, and service accounts with permanent keys [2] for use with CI servers etc. I find both of these to be fairly straightforward to use.
