Streaminsight - Query around set-up and hardware (beginner)

hi.

looking using streaminsight on project @ work , have read lot of articles i've not found details around hardware requirements this. have seen minimium set-up documents, sort of after real life examples.

for example if trying process events on hour between 10,000 , 20,000 events. keeping in memory process - take? if on day.

is there considerations when server fails? can recover of events (expect re-processing) or mirroring server?

sorry if these aren't technically correct, getting head around concepts.

thanks in advance

ian

actually, don't have add powertx125 said .. he's absolutely correct. 10k-20k events/hour that's pretty easy streaminsight handle. how cpu need depends on complexity of analytics , how many output sinks have. how memory need depends on of same payload size. said, streaminsight pretty efficient when comes memory usage.

as raw throughput ... little update txpower's numbers ... i've gotten 150k events/second throughput on quad core i7 laptop passthrough query writing local hadoop instance. there basic analytics on calculate events/second , write them console minimal. cpu utilization hovered around 75-80% there no significant output queue - no more 1-2 second's worth of events @ 1 time.

in real world scenario, complex analytics, we've seen on 100k events/second on dual quad core xeon 32 gb of ram. , average of 35% cpu.

---

devbiker (aka j sawyer)
microsoft mvp - sql server (streaminsight)

---

---

if answered question, please mark answer.
if post helpful, please mark helpful.

SQL Server  >  Microsoft StreamInsight