Platform Maturity Testing, Jmeter Single VM, 64GB RAM, Dual hex core 3.3ghz xeon cpus
| Test | Result | Comments |
|---|---|---|
GET ALL 10 threads alternating 220k (models) and 6k (named-queries) json payloads. | Generate Summary Results = 17424 in 603s = 28.9/s Avg: 344 Min: 68 Max: 1198 Err: 7 (0.04%) | |
PUT customer + allocated-resource + cross-connect + service-subscription + service-instance GET customer DELETE customer 10 threads 1000 loops, 10000 customers built and torn down | Generate Summary Results = 30000 in 00:13:25 = 37.3/s Avg: 267 Min: 7 Max: 3110 Err: 0 (0.00%) | |
PUT customer + allocated-resource + cross-connect + service-subscription + service-instance GET customer DELETE customer GET ALL 220k payload on models 20 threads 10000 loops, 200000 customers built and torn down | Partial progress: Generate Summary Results = 156377 in 01:25:36 = 30.4/s Avg: 656 Min: 20 Max: 3729 Err: 0 (0.00%) After 300k requests, starting seeing errors in cassandra: org.apache.cassandra.db.filter.TombstoneOverwhelmingException. Investigating. Too many deletes with too long a grace period caused the tombstone exception cqlsh> use aaigraph; cqlsh:aaigraph> alter table graphindex with GC_GRACE_SECONDS = 3600; Retrying test. Retried: Generate Summary Results = 800000 in 07:56:02 = 28.0/s Avg: 713 Min: 18 Max: 4455 Err: 0 (0.00%)
|
