We investigate the creation of cold dark matter (CCDM) cosmology as an alternative to explain the cosmic acceleration. Particular attention is given to the evolution of density perturbations and constraints coming from recent observations. By assuming negligible effective sound speed we compare CCDM predictions with redshift-space-distortion based f(z) sigma_8(z) measurements. We identify a subtle issue associated with which contribution in the density contrast should be used in this test and then show that the CCDM results are the same as those obtained with LambdaCDM. These results are then contrasted with the ones obtained at the background level. For the background tests we have used type Ia supernovae data (Union 2.1 compilation) in combination with baryonic acoustic oscillations and cosmic microwave background observations and also measurements of the Hubble parameter at different redshifts. As a consequence of the studies we have performed at both the background and perturbation levels, we explicitly show that CCDM is observationally degenerate with respect to LambdaCDM (dark degeneracy). The need to overcome the lack of a fundamental microscopic basis for the CCDM is the major challenge for this kind of model.