lzimm/qluster
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
-- this is fairly old school work yo -- so, after i did that vurs thing (ie: http://github.com/lzimm/vurs), i had this cluster framework i started working on that i figured even though was full of funny stuff all over the place (cuz it did really silly things, like rely on ip broadcasts to maintain heartbeats of all the nodes), and didn't do anything in terms of caching or anything like that, it had enough put together that i figured "hey, even though things like that are a pain in the ass, (because of all the fun issues that creep up in invalidating things properly in the middle of data races happening all over the place, and balancing between stampeding and serving what could be dangerously stale data), this stuff is layered enough that i can probably get to it pretty easy if it ever becomes that painful of an issue" (and i'm talking about lazy caching, where it just does it all magically so entity reads stay nice and snappy, regardless of whether that's even a good idea in the first place, that's for the scientists to figure out). besides, even though not much of it was very clever, it had what, at the time, felt like a fairly nice serialization system, so that nodes could communicate pretty easily (in a consistently hashful way), had a "nice" modular entity system that i could use to swap out storage backends for different types of objects between mysql, cassandra, hbase, binaries, and whatever else i wanted to "nicely wrap" in the future, and it even had this little multi-layer cluster ring system, where, since every node was homogeneous, i could set them up to participate in certain subsets of "service layers", with independent capacity requirements, and by monitoring usage and all that jazz, i could automatically balance between them, even though leaders on said "layers" were "elected" in a way that i wouldn't exactly back up in a knife fight. what a magical pipedream. plus it did all this stuff to do dependency resolution between internal services (in what can actually be best described as "keep running against the wall until you break through it"), and, since i basically ripped it out of that vurs thing, it had all this strange NLP stuff (which i only pretended to understand) baked in to boot (for better or worse, whether i wanted it or not). but yeah, as you can probably tell from my tone, i don't touch this thing anymore. anyways, standard disclaimer: there's a bunch of extra junk in here you probably don't need to run it, while on the other hand, its probably missing some stuff you do (as well as any explanation of how all this junk works, license information that i probably stripped from a bunch of files and any sort of sanity). so funny.
About
java based cluster "framework" i used on a couple projects, for no reason.
Resources
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published