Product
Support
Everything Else
Helix Server 6.0 & 6.1 Performance Testing
Performance Tests

Ad hoc performance comparison: Helix Server 6.0 vs. 6.1.

  • Servers:
    • iMac: 1.8GHz G5 w/512MB RAM running macOS 10.4.11
    • Mac mini: 1.83GHz Core 2 Duo w/3GB RAM running macOS 10.5.2
  • Clients:
    • Client 1: G4/733 w/1.5GB RAM running macOS 10.4.11 + Classic 9.2.2
    • Client 2: G4/450 Dual w/2.0GB RAM running macOS 10.4.11 + Classic 9.2.2

Missing numbers indicate either test not done or unable to complete.

Test Stage iMac G5 1.8GHz Mac mini Core 2 Duo 1.83GHz
Server 6.0 PowerPC Server 6.1 PowerPC Server 6.1 PowerPC (Rosetta) Server 6.1 Intel Native
Small Data w/Count

Client 1: Posting 13 iterations (from 2^0 - 2^13) of small records
Client 2: Idle, with an entry view open displaying the record count only.

Posting Done 1:00 0.58 1:06 0:35
Quiescence 1:14 1:09 1:20 0:38
Large Data w/Count

Client 1: Posting 13 iterations (from 2^0 - 2^13) of large records (1 each large fixed point, styled text, plain text, doc, & pict)
Client 2: Idle, with an entry view open displaying the record count only.

Posting Done - - - -
Quiescence 1:24 1:20 - -
Large Data w/List

Client 1: Posting 13 iterations (from 2^0 - 2^13) of records (1 each large fixed point, styled text, plain text, doc, & pict)
Client 2: Idle, with an entry view open displaying the record count and a list view open displaying the records as a list.

Posting Done 1:16 1:12 - 0:55
Quiescence 6:20 5:28 - 3:58
Notes:
  • 2^0=1. 2^1=2. 2^2=4. 2^3=8... 2^13=8192. Total number of records posted = 16383.
  • Posting Done marks the elapsed time from the moment the sequence started until the final posting dialog disappeared on the Client's screen. This is effectively the amount of time it takes the Client to send, and the Server to process, those requests.
  • Quiescence (a state of inactivity or dormancy) marks the elapsed time from the moment the sequence started until all Clients had caught up with the changes and the Server was idle again. This is the time between when the Server finished processing the user's actual data and the time it finished notifying all of the connected Clients about the changes that were made. More Clients means more messages sent out means a longer time to quiescence. (Notice that when Client 2 has a list open, the time to quiescence is much longer than when it only has an entry view with the recourd count open.)
  • A quick test with an Intel-native Client running on a 2.0GHz Core Duo MacBook w/2GB RAM running macOS 10.5.2 showed promise of more gains.