Hi there,
one approach you could use is to separate your test in two scenarios, each with 50 VUs, with one scenario running first and the other one after, using the startTime option.
Selecting a unique JSON object per VU is a bit tricky, but doable with this approach.
However, selecting objects sequentially is even trickier since you can run the test with arbitrary VUs and iterations, and you’d need some centralized way to increment the index. For this I’d recommend considering the xk6-counter extension, for which you’ll need to compile your own k6 binary using xk6. Otherwise you could also use Redis for this purpose, using one of the Redis extensions, or even over HTTP with Webdis. You’ll need to use the Redis INCR command, but it should be relatively straightforward.