dastbe

17 ms·

151.
by dastbe 5y ago
Does anyone have more context on why “unable to retrieve their xDS configurations.” led to ” Traffic Director-managed clients deprogrammed as the configuration was removed” ? For example, Envoy will keep its configuration forever unless 1)
152.
by dastbe 5y ago
> Scalable clouds are 100s to 1000s of times more expensive than using your own hardware. This is trending into hyperbole. a c6g.metal at AWS is $21,502 for a 3 year all upfront reserved instance. You're telling me you can pay just
153.
by dastbe 5y ago
i meant with. google has built some really great tools for minimizing toil when pushing major changes across a codebase. even then, there’s a nontrivial overhead that is borne across the company. > Why would you say that is? many/mo
154.
by dastbe 5y ago
if you have the right company setup, headcount, and culture then you can absolutely build a world where teams can be allowed to force “perfection” over time, even with amazing levels of automation. i think the interesting thing is that whil
155.
by dastbe 5y ago
building a public cloud is more or less a rebuild of any infrastructure you currently have. it would be just unwise for a company to build their private cloud with the same trade offs as public cloud. so while there’s the ideal state of run
156.
by dastbe 5y ago
while i'm sure they could get something out the door in 6-12 months that would satisfy somebody somewhere, I'd say more like 3+ years before you'd have anything that any companies would start betting the farm on.
157.
by dastbe 5y ago
> AWS started because Amazon wanted to lease out their spare infrastructure capacity. No, it really did not. The reality is even if Meta has the best data center infrastructure in the world their infrastructure is focused solely on solvi
158.
by dastbe 5y ago
fwiw i think you’re both saying something very similar. true time has to be correct about max skew in order for it to not break the assumption spanner is built upon. you could also use a looser time bound and still have correctness, but end
159.
by dastbe 5y ago
while that could be preferred, here's some reasons why it may not be 1. pre-existing behavior. maybe queries have only started getting big enough to where this mattered, but all of our existing clients communicate the original way and
160.
by dastbe 5y ago
I used to work at Amazon, but even then it's pretty clear the support person is erring on the side of caution here. in isolation two grocery store addresses and orders isn't particularly illuminating, but tying it with their real
161.
by dastbe 5y ago
advisory locks sound interesting, but this really sounds like a problem that the author is forcing on themselves. i get the benefit of having your database persist the immutable ledger, but it seems like you would have a much more scalable
162.
by dastbe 5y ago
if you think of paxos/raft as having fault tolerance of losing n nodes, then the "decide what happens" part of 2 phase commit can be thought of as distributed consensus with a fault tolerance of 0 nodes. iirc leslie lamport h
163.
by dastbe 5y ago
> Amazon seeks to avoid common-mode failures between AZs (and thus regions). there is a distinction between azs within a region vs azs in different regions. the overwhelming majority of services are offered regionally and provide slas at
164.
by dastbe 5y ago
how many rsus do you have depending on it
165.
by dastbe 5y ago
i would guess as a percentage there are two orders of magnitude more senior engineers at an arbitrary tech company vs managing directors at an arbitrary investment bank. senior engineer is ~5 years of experience to achieve, with maybe 1-2 y
166.
by dastbe 5y ago
to be clear i don't think you should mind if comp-maximization isn't your sole focus now (or ever).
167.
by dastbe 5y ago
It is an unfortunate truth that Microsoft's pay between entry level to somewhere around principal is not competitive with a substantial number of companies. That is not to say that Microsoft doesn't provide other benefits, but top
168.
by dastbe 5y ago
are you interested in the norm in terms of raw numbers, or where you would need to work to achieve this? and are you interested in people who have won the stock lottery, or just people who have their total comp pegged at 500k+? assuming the
169.
by dastbe 5y ago
> It's just the in-app premium monthly membership fee we're discussing? Sure, but even then Stripe charges half a percent extra to handle subscriptions. Given Uber One is 10$, my original assessment still stands that Apple want
170.
by dastbe 5y ago
should apple be allowed to make a (what seems like substantial) profit off that lock-in? by comparison payment processors like stripe can offer as a baseline 2.9% and 30 cents. if you predict that the average uber is 10 dollars (which is pr
171.
by dastbe 5y ago
anecdotally i saw a huge dip in consistency across both ubereats and doordash in the back half of 2021. like orders taking 1+ hours greater than expected delivery. that being said, i've never had the problems you're citing with ei
172.
by dastbe 5y ago
are you not involved in the product roadmap and how that gets translated into the individual, smaller projects that get you there? then you may not be as senior as you think ¯\_(ツ)_/¯
173.
by dastbe 5y ago
they do. the docs are open source and you can propose changes to them https://github.com/awsdocs
174.
by dastbe 5y ago
because it’s a complex product, and having empathy for customers is far more helpful than having contempt.
175.
by dastbe 5y ago
if you read my first comment i never argued for getting rid of the sat, though they should get rid of the usurious prices for tests and sending your scores out. in fact, i argued that of all the subjective tests the sat has the most legible
176.
by dastbe 5y ago
that would work if you were talking about random people who were selected to do a surprise track race, but if the track race is known to be THE measure and that can be influenced by training then you’re going to see skew. to be clear, i sti
177.
by dastbe 5y ago
subjective/objective are probably not the right adjectives here, but to use your example I would say that a timed foot race is an objective measure of how well you do in timed foot races, but is not an objective measure of your innate
178.
by dastbe 5y ago
put a big IMO/IMU around all of this, but the SAT/ACT isn't objective in the sense that studying for them is something you can do to improve your score and your prospects. However, it is also transparent in how you do that: y
179.
by dastbe 5y ago
> There are rare technical reasons to choose micro services Is this just the difference between working in an infrastructure oriented space vs. product oriented? in infrastructure i find that it is often the case that most logical applic
180.
by dastbe 5y ago
I'm curious, has anyone leveraged bazel's package visibility rules to improve isolation within a monolith? one of the things i don't like about monoliths is that tight coupling is often silent rather than noisy, whereas if so

More ›