The claim "that a deadlock in and of itself is not necessarily a problem" and suggestion that "A properly designed and coded application ... can intercept the 1205 error and resubmit the deadlock victim request back to SQL Server" is forgetting a few basic things. Imagine, for example, if all client applications to any server they talk to had to retry their requests all the time - like, say, your web browser to any web server? What was rather meant, I hope, is that a properly designed database would never get you into unexpected deadlocks.