More important than LIRP influence is that Bay Area tech companies were incredibly successful in the decade at increasing revenues, so they could afford to pay more.
Apple: $65.2B in 2010 vs $260B in 2019
Facebook: $1.97B in 2010 vs $70.9B in 2019
Netflix: $1.67B in 2010 vs $20B in 2019
Google: $29.3B in 2010 vs $160B in 2019
IoT VLAN indeed can be annoying. It's getting better as a lot of the more "prosumer" grade routers are supporting it. I use Sonos at home too, which means I had to deploy this into a VM to bridge the VLANs: https://github.com/alsmith/multicast-relay.
There are some funny (?) things that turn up too, like learning the Roku remote iOS app "discovers" devices by opening a TCP connection to every address in parallel on its local /24 (!!!). It sends out and receives mDNS packets that would tell it exactly where they are, but they are ignored by the app.
Everything has a purpose, unlike many "home labs" where people are just tinkering. There's nothing in here that would require fussy maintenance. It seems pretty reasonable to me given the functionality.
It didn't impact everything, I think I was hitting this case:
> When a query is deemed too complex to determine what specific cached data may have become invalidated, a fallback to a simple but effective table level invalidation occurs.
I tried to use PolyScale in the past but had issues with performance because updating a row would invalidate the entire cache. I wonder if that has improved?
It was removed from MySQL in 8.0 because it wasn't very useful. MySQL query caching does exact matching on the query string and any row update to a table used for a given cached query nukes the entire cache. So it's only useful for a small set of niche use cases where tables are essentially static.
What is fundamentally concerning about a central bank injecting capital into the system? If anything, you could say the same thing about fiat currency and fractional reserve banking as a whole. Money comes from nowhere!
What's the core ordering mechanism behind the serializability for arbitrary distributed transactions? This would be something like TrueTime in Spanner or FoundationDB's GRV/commit proxies.
It feels like a lot of money, but I'd point out that even venture's blockbuster 2021 year, where it raised ~$128B in funding, is still far less money than Apple's ~$202B and Google's ~$169B cash on hand. Those are single companies.
I love all the nostalgia, but the post doesn't really answer the most interesting part of the title: why do workstations matter? I was really hoping there was some revelation in there!
You can't share objects across threads (called "processes") in BEAM, so it's very different. The GC only ever needs to pause one call stack at a time to do a full GC cycle. Memory shared across processes is generally manually managed, typically more akin to a database than an object heap.
This outcome should also be entirely unsurprising and should pass people's basic sniff tests as Timescale works within the existing, mature architecture of PostgreSQL, where-as ClickHouse is a greenfield single-purpose system. Software makes trade-offs.
That would be rare, because rapidly increasing property values are almost always driven by rapidly increased demand. The municipal government has to expand its expenditures to meet this demand, usually ahead of when it receives expanded revenue. Generally speaking, it's much more expensive to grow quickly than slowly.
It does. The three AZs they replicate across are just as good as anything else someone typically calls a "datacenter." Amazon itself operates retail out of one region and uses multi-AZ as a fault tolerance strategy.
That's an apples to oranges comparison though. Upstash couples durability with consistency/isolation. Regardless of configuration, FaunaDB and DynamoDB both always ensure durability of acknowledged writes with a fault tolerance of greater than 1 node failure. To compare them on equal footing, Upstash would need to be configured for strong consistency, at least according to the docs.
Yes, even for on-disk read-only use cases, shapefiles are almost always much faster. Overhead from serialization, network, etc often dominates the performance of GIS.
How would you expect that to actually work though? If a goroutine could be arbitrarily stopped at any point, that would trivially lead to degenerate program states. A lock could be held open or another goroutine could be waiting on it for a message.