Cloud tools / performance vs. EXAoperation

drumcircle
Rising Star

I'm observing that the Cloud Tools UI is taking a very long time to perform resize operations, more than 1 hour to add 2 nodes to a test cluster for example.  Is cluster resizing and instance resizing actually any slower than the "old fashioned" way?

1 ACCEPTED SOLUTION

drumcircle
Rising Star

It has been suggested that CloudTools automatically runs a reorganize command on the cluster, which is an expensive operation.  It's not clear if scaling up/down CPU only, also triggers reorganize.

View solution in original post

5 REPLIES 5

mwellbro
Xpert

Hi @drumcircle ,

not that I´d remember it to be significantly slower....

Could you give a few more details: manage- and data node instance class , region and az ( I´m again assuming AWS - if not: which provider ? ).
Also: how big in terms of storage is your cluster at the point of the resize ? And what kind of storage is it ? ( gp2 or hdd or X ? )

Since those were probably on demand instances: did you check "VM provisioning time" ?

Cheers,
Malte

drumcircle
Rising Star

AWS upgrade of 2 nodes from m5a.4xlarge to m5a.16xlarge took almost 2 hours with Cloud Tools.

I've done instance upgrades like that in < 15 minutes without.  There is a BigTable parameter that is tricky so I like using Cloud Tools if I want more memory available, but often instances are CPU not RAM limited.

 

Perhaps yesterday was a bad day in US-EAST-2

exa-Chris
Community Manager
Community Manager
Hi @drumcircle, let me have one of the cloud engineers have a look into this. As we are heading into the Easter Weekend, that might take till Tuesday if that is OK. Rgds Christian
Connecting Customers, Partners, Prospects and Exasolians is my passion. Apart from that I cycle, listen to music, and try to understand what all those technical discussions really mean...

drumcircle
Rising Star

Sure thing.  I had to restart EXASolution and CloudTools a few times yesterday.  This is not constructive feedback per se and not a behavior I'm accustomed to, perhaps my install is foobar somehow.  But no data corruption to note and no reduction in run-time query performance.

drumcircle
Rising Star

It has been suggested that CloudTools automatically runs a reorganize command on the cluster, which is an expensive operation.  It's not clear if scaling up/down CPU only, also triggers reorganize.

View solution in original post