back
5 comments
Maintain large kafka clusters, futz with mesos, and futz with kubernetes. Same as all of us really. The difference is that a lot of startups will talk about their "hard big data problems" and might have 500T. A firm like Bloomberg's "big data problems" might be 400+ Petabytes. They also generally speaking totally shun the cloud as at that scale it tends to be cheaper to build on-premise.
They also shun the cloud because they have one of the largest physical datacenters on the island with real heavy iron (IBM and Sun/Oracle boxes)
There was a talk at the recent (quadrennial) DTrace conference from someone at Bloomberg which gives some numbers instructive of what at least some staff in tech there do: https://www.joyent.com/about/events/2016/dtrace-conf/videos/...
In 2014 Bloomberg had 19k employees and 9.3B turn-over. There must be something going on there[1].

[1] http://www.forbes.com/companies/bloomberg/

Bloomberg has some of the best tech departments I've had the pleasure interviewing with. Besides being a media company, they also run a few exchanges, such as one of the largest bond exchanges in the world, and their tech people are on point. Totally to rate. I completely underestimated them until the recruiter I was working with set me up with them, and I was completely shocked at how good they were. Their top group at least was very very talented.
Sell those terminals and connections that are tens of thousands of dollars per year (per month?).
For Bloomberg, it is $20,000 USD per user per year. One user == one traders. Most firms have more than one trader.
It's more than just traders. On the sell side, you'd also see salespeople with terminals (and often the middle office guys would also have a terminal).