Redshift Research Project

Blog

The ds2 node types have been retired in regions af-south-1, ap-east-1, ap-northeast-2, ap-northeast-3, ap-south-1, ap-southeast-1, ca-central-1, eu-north-1, eu-south-1, eu-west-2, eu-west-3, me-south-1, sa-east-1, and us-west-1.
The ds2 node types remain available in ap-northeast-1, ap-southeast-2, eu-central-1, eu-west-1, us-east-1, us-east-2, and us-west-2.
The dc2.large node type in eu-west-2 has lost the disk read benchmark improvement (and so has returned the 0.14s, rather than 0.06s).
A few regions last benchmark were having a bad day, with unusually slow results. Those regions have come back to the fold.

ds2 types have now been retired from all regions.
It’s apparent some regions, this only for the dc2.large type, can be described as having, for the network benchmark, a “fast” mode or a “slow” mode.

ap-northeast-2 network benchmark for dc2.large is, over the course of all benchmarks made so far, variable. The sequence of timings, starting from this most recent benchmark, are : 7.33s, 7.94s, 4.72s, 5.02s, 9.65s, 9.22s, 5.27s, 5.53s, 8.40s, 8.99s, 8.26s.

ap-northeast-3 shows similar but more extreme behaviour : 13.26s, 13.54s, 13.37s, 5.12s, 14.38s, 5.47s, 4.99s, 5.99s, 7.44s, 6.94s, 7.19s - which is to say, for the last month and a half, the network benchmark in that region is taking about 2.5 times longer to complete.

ap-southeast-1 has something of the same, normally being about 5s, but for the last two months being 8s.

eu-north-1 for the last 3 months is at about 13.5s, then had one benchmark of 5s, then another month of 14s, then 1.5 months of 5s.

All other regions have been consistent over the last 5.5 months of benchmarking.
eu-west-2 dc2.large nodes have received the doubling of disk read performance seen turning up in most regions (albeit in some regions sometimes present, and then not present, and then present again).
us-west-2a was out of capacity for ra3.xlplus, so there’s no benchmark for that node type in that region.