What’s new with our in-house load balancing service
dropbox.tech
dropbox.tech
Why use both?
Why even consider ZK in 2024? Is it just popular in java shops?
Maybe I am misunderstanding?
I'm no expert on these consensus-stores, but I guess my question would be "why not?" ZK is proven and works.
Cool to see some other approaches, although I believe they've complicated theirs a bit more than necessary.
I heard about someone using PO2C for cache eviction recently. Maybe Oxide talking about their storage work? Seems obviously smart. No need to keep LRU or carefully track & decide, just use statistics to compare two random elements and drop the least used one.
According to his LinkedIn profile: 7 years 6 months into the job.
Although, it looks like he transfered to a new team a year ago to do ML Ops.
That’s the only reason they named the software “Robinhood”.
Why not Robin Hood or some other unique or generic name?
Context matters. There simply isn't a risk for meaningful confusion here. So what's the issue exactly?
I don't know about that. Why do you think people would not link a financial company to the folklore character famous for redistributing money? That's kind of disingenuous claim to me.
When I first heard about it, I actually assumed it was a reverse Robin Hood by taking money from the poor to give back to the rich type of scheme.
That space or the lack thereof is important.
To date I only know of two Robinhoods, the brokerage and the load balancer, and I only learned about the second with this thread.
In fairness the load balancer presumably came first, but the point remains.
Or maybe not, try actually naming something sometime and you'll see. It's not easy to come up with something that fits all the requirements AND has never been used before.
And it's not like anyone is ever going to get confused and accidentally try and use Dropbox's internal load balancer to buy options on a stock or something. Or try and make a gambling app the star of their next hit movie for that matter.
Imagine knowing that Robinhood exists as one of the most popular (or the only) policy engines for large (EB sized) filesystems for but still naming your new company Robinhood.
I think Robinhood Markets should C&D immediately and find themselves a new name.
Quite insufferable if you ask me.
I have computers named after stars and galaxies. Some codenames of the applications I have developed myself has stupid connections to characters because of what they (the software) do are similar to characters themselves. I named a module "Scrat" because it was a state machine which changed thousands of states per second according to input it received from a file I/O module called Manfred (which carries things on his back or with its trunk).
I have my names clash with others, and others' names clash with mine. Who cares, it's fun and makes things enjoyable for everyone, plus it allows one to inspire themselves by setting a vision for their software.
Or, maybe we should start naming our software with UUID strings. That won't clash in a long time. :D
Heck even code does scoped naming... the compiler can't really handle punctuation that is slightly wrong and even it allows me to reuse variable names in the same program.
I don't know why we'd need a univerally unique name for each piece of software ever created. The existing software called robinhood is unrelated to this load balancer, so even if I could download and use it, I doubt I'd have much of a problem distinguishing the two.