Considering the fault of any experimental evidence from LHC for supporting the supersymmetry idea until now, can we say that it is dead? Generally the people who are working on this subject say that MSSM probably is dead but we have some new extensions of it such as NMSSM. This argument seems weird a bit. Sometimes I feel that it is a vain try to keep an idea live because they have invested their scientific life on it. On the other side, I have to confess that the supersymmetry is a beautiful idea. But it seems that the nature does not like it and we have to accept it. Other related question is, if the supersymmetry fails, can we say that the superstring fails too?
No, we can't deduce anything about the correctness of supersymmetry as a principle from the LHC data available so far. There exist supersymmetric models that are compatible with everything we've seen; and there exist non-supersymmetric models, especially the Standard Model as the minimal one, that agree with everything so far, too. So there's no way to empirically discriminate.
The only way to decide such Yes/No questions in science is falsification and the class of supersymmetric theories has not been falsified. At most, a fraction comparable to 3/4 of its parameter spaces have been ruled out. The right point of the parameter space may be in the remaining 1/4. It is not an "unlikely" assumption in any scientific sense. If the fractions of the volumes are counted as probabilities, the refusal of SUSY to show up so far may be quantified at most as a 1.5-sigma "bump" testifying against SUSY which is a negligible amount of evidence relatively to other evidence we have.
Now, it is also untrue that the MSSM is dead. At most, some heavily constrained "special cases" of MSSM such as CMSSM (essentially the same thing as mSUGRA for these purposes) may be nearly dead now. But others, such as pMSSM which is an analogous and arguably better motivatived subset of MSSM, are alive.
The validity of string theory is independent of the appearance of SUSY at the LHC or any collider in a foreseeable future because there exist string vacua – and, similarly, regions in parameter spaces of effective field theories – in which SUSY is broken so that pure Standard Model is left up to rather high energies.
Now, it's also preposterous to call NMSSM a "new extension". The NMSSM goes back to 1975 so it's almost as old as SUSY itself, see the 1975 paper by Fayet and much more detailed supersymmetric papers from the 1980s in reference 4 here which established much of the physics of the NMSSM. Even when we demand some details, the NMSSM has been a known candidate theory for 30 years.
More generally, all the suggestions that SUSY theorists are moving their SUSY models towards contrived, high-energy versions are mostly fantasies. Since the beginning, people would talk about superpartners in the range of hundreds of GeVs to few TeVs and at this qualitative level, this is true for the papers describing currently viable models. Also, one may see that the SSC that was supposed to probe new physics of SUSY and similar things had the center-of-mass energy 40 TeV, five times higher than what the LHC has shown so far. So at the energy scale, we're still well below what the physicists found appropriate to "reliably enough" find SUSY 20 years ago. Some people would argue (without evidence) that 7 or 8 or 13 or 14 TeV could be enough to safely see "everything" as well but at least in the case of 7 and 8, it is evidently not the case.
Concerning "invested lives", SUSY research is alive and kicking and SUSY remains the #1 scenario for new physics to be discovered. In fact, its relative weight is getting larger, not lower, because all other ideas' parameter spaces are being more quickly ruled out by the experiments than SUSY's. It's also untrue that SUSY is being worked on by "people who have invested something". Most of the SUSY papers are being written or co-written by young people who have gotten started with physics in recent 5 years.
To summarize, all your "feelings" against SUSY are entirely unsubstantiated, based on faulty data and invalid logic.