Imported from previous forum
[ original email was from Steve Glickstein - sglickstein@warburg.com ]
Can anyone share some test scenarios they would recommend performing before bring a new fix engine/application live?
[ original email was from Witold Sames - wsames@bloomberg.net ]
> Can anyone share some test scenarios they would recommend performing before bring a new fix engine/application live?
> From my experience, areas to concentrate on are:
- Proper handling of PossDupes (BEHIND the FIX engine itself)
- Accurate Cancel_Replace mechanism such that blotters agree on the status of an order, OrderID incrementation, and the like
- SeqNum reset scenarios such as resetting to valid and invalid numbers and appropriate reaction of your engine
- Custom field handling and mirroring if needed
- Recovery from errors
- Reject handling
For more details on how we test a FIX implementation, you can contact me directly.
I concur with the above with the following additions:
- Price and/or Quantity corrections on reports
- Corrections of corrections
- Corrections while in Cancel Pending or Replace Pending
- Busted reports on partially completed orders
- Busted reports on completed orders
- All of the above scenarios after one or more Cancel Replaces of an order
Obviously these are business cases we hope we don’t have to deal with often, but we see corrections from DOT and ECN’s daily - sometimes well after the original report. As the first respondent indicated, do not overlook the “unusual” comm problems (restarts, seq resets, etc.), but also cover all the “real world” business issues - at least to have seen many of those communications and business situations before you see them in production.
For more information on the testing CIS performs and our test facilities, please contact me directly.