In optimization, one often has to solve a symmetric indefinite system. Optimality conditions depend on the inertia of this system. This talk covers a modified LBL^T factorization which allows for a transparent revealing of the matrix's inertia and modification thereof.