@MarcJBrooker Do I read it correctly? At 0.15% (15 rows out of 10k) the cost of acquiring row locks reach the same amount as that of the table lock at read. Or should I read it like acquiring 15 row locks costs about the same as read-locking the entire table , no matter the size.
I am obligated to mute @elonmusk because his excellency's egomaniac change request made my timeline useless. No, Elno, everybody is not interested in everything you say.
@amdev@maybettl@bidetofevil@ZoeSchiffer Salvage some hw from smf to be shipped to pdx should be the plan. Feasible but not reasonble. Wonder if he actually understands how important a role the site failover plays for the reliability of Twitter and what peak failover implies. Does he know load shedding is not there yet
@Carnage4Life Does the WARN act protect those that are on work permits? When does the timer to find another job to transfer the work permit start? As far as I understand the legislation is not clear. So many tweeps chose to err on the side of caution.
@Carnage4Life What should prevent a CEO to treat thousands of employees as merely office furniture? IMHO the CEOs sriamk mentions wouldn’t be that excited if the immigration limitation didn’t hold back the impact of resignations in Twitter.
@imightbemary virtual desktops that contain those dashboards. it catches your eye when you switch left & right. :P Seriously a regular, dedicated team review meeting on a reasonable cadence + alerting was sufficient. IMHO big TVs displaying those dashboards are only fit for a command center
@Jason It is this very same genius who made the back of the envelope calculation to conclude that twitter’s problem was the size of its workforce rather than unsuccessful execution leading to insufficient revenue so 75% layoff would fix it.
@Jason Twitter 1.0 was in the first category. 2.0 is in the last one. We are yet to see the productivity levels of 2.0. So far it’s exhausted employees despite nothing significant being shipped.
@rynonl@mssoran No, TSDB is after Manhattan. It is a nosql db built to store time series data efficiently. Cuckoo is still there but it changed a lot, as you might have guessed. Viz is no more. We built a newer stack instead and migrated all the dashboard and alert configs.
During the last 4 years, the observability data grew by 10x. We scaled our services to meet the demand. That's the easy part. We did it while keeping the costs the same. Now, **that** is not so easy. That's what those SWEs were for. #Observability#TwitterPlatrofm