Hi Group,
I am trying to size and create my ec2 cluster. My index size across all
cores is approximately 25gb. I have experimented with configurations using
as little as 512M and as high as 32gb. My problem is that when I keep all
things equal my qtimes are all over the place. All machines are using 3000
IOPS 125 MB throughput gp3 EBS, except where noted. One machine is using
5000 IOPS with 200 MB throughput
For instance:
t4gsmall - 139,
t4gsmall 2 - 2320,
r7gxl - 4841,
t4lg - 3183,
t4lg (with iops5000 gp3 ebs) - 1757,
t3.large - 266,
my localmachine - 388,
Some additional information. I created an AMI from the first t4gsmall above
and the next 4 were created from the AMI. The second to last one was a
different aws account where we were beta testing some things, and the final
one is our local dev vm, a 12 year old HP server.
I'm just perplexed, particularly with the first two on the list. The server
specs are the same, the configuration identical. We are using the oler solr
7.7.3, but I don't see why that matters.
Does anyone have any ideas? I'm not sure why these QTimes are all over the
place. I have performed dozen's of queries and the breakdown is always
about the same.
Best,
Matt
