I'm testing out some BigTable queries on a 3-node cluster using the Go client, like:
r, err = tbl.ReadRow(ctx, "key1")
I'm getting results back within a few ms:
query 1: 129.748451ms
query 2: 3.256158ms
query 3: 2.474257ms
query 4: 2.814601ms
query 5: 2.850737ms
As you can see there's a significant setup connection delay on the first query. Can anyone provide feedback whether this would be an acceptable value? The queries originate from a GCE VM in the same zone (europe-west1-c) as the BigTable cluster.
Furthermore, is there any support planned to pool the BigTable connections when running on App Engine?