You could use MayaChemTools for similarity based virtual screening using 2D fingerprints. Comparing your molecule to datasets like ZINC, ChEMBL, Drugbank, etc. In this case all you need is Perl installed and a couple of CSV files, preferably with "curated" records. For commercial options Canvas or MOE come to mind.
For 3D comparisons you could use USR-VS:
http://usr.marseille.inserm.fr/
This server needs a .sdf file as query and searches similarities to it on ZINC database. Thus is more straightforward, however the active (or presumably active) conformation of the molecule should be used for better results.
Of course there are more nuts and bolts to it, but this may help you on your search.
Consider the following reference for further reading:
I agree with above researchers as virtual screening is method for this. But the simplest way is to search your molecule in Pubchem and then find the similar molecules given in the pubchem database like similar physico chemical properties, seed molecule etc.
It will be appropriate if you go for ligand based virtual screening which will ultimately give you n number of structurally similar molecules on the basis of lipinski rule as deposited in the respective databases.
Simply get the SMILES format of your initial molecule from the design software if it is novel. But publicly available, you can get that from pubchem compound. Then use that SMILES to query Swiss similarity server (www.swisssimilarity.ch) by choosing the right drug database for each run.
You will get multiples of similarly molecules ranked by scores with link to drugbank information.
Here, good thing is as mention above reply via Dr. Toluwase H. Fatoki, you can use ZINC database as well as ChEMBL database also. And you can get that from Pubchem compound database also. Here in Pubchem database you can use molecular formula and screen the compound as selected percentage cutoff as 80%, 90%, or 95% similarity likewise.