Skip to content

Commit e981f88

Browse files
committed
more stuff
1 parent 2e1de00 commit e981f88

File tree

1 file changed

+10
-0
lines changed

1 file changed

+10
-0
lines changed

doc/developer/design/20250717_a_small_coordinator_more_scalable_isolated_materialize.md

Lines changed: 10 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -50,6 +50,16 @@ Another way of seeing this in action is of course when one use-case is using
5050
the environment a lot, say high SELECT throughput: this will also make the
5151
system less responsive for other use cases/users of the system.
5252

53+
Both of these _might_ not obviously be problematic today, but we think we will
54+
have customers soon for which our scalability limits become a blocker for
55+
expansion and we do see a trickle of bugs where the lack of isolation makes it
56+
so that Materialize becomes unresponsive for customers, for multiple seconds at
57+
a time. This can shake confidence in Materialize, which would also be a blocker
58+
for adoption or expansion. There is some recency bias to thinking how urgend
59+
these bugs are, but here's the latest example, where we had to disable a
60+
feature because it can block the coordinator main loop for 10s of seconds or
61+
more: https://github.com/MaterializeInc/materialize/pull/33070.
62+
5363
## Success Criteria
5464

5565
We want to address both of the problems mentioned above, but scalability is the

0 commit comments

Comments
 (0)