This paper investigates the intricate relationship between the topological properties of the set of invariant subspaces of a linear operator and the operator's canonical decomposition. The set of all k-dimensional subspaces of a vector space forms a Grassmann manifold, which we endow with a natural topology induced by metrics such as the gap metric. We explore how the continuity and stability of an operator's invariant subspaces under this topology are directly linked to the structure of its Jordan or Rational Canonical Form. The analysis reveals that the clustering of eigenvalues and the sizes of Jordan blocks, or the structure of invariant factors in the rational form, determine the topological characteristics of the corresponding subspace lattices. Specifically, operators with well-separated eigenvalues exhibit locally stable invariant subspaces, whereas those with clustered eigenvalues or large nilpotent blocks lead to more complex and sensitive topological structures. The paper demonstrates that the canonical decomposition provides a fundamental algebraic framework for understanding the geometric and topological behavior of invariant subspaces under perturbation. This connection is crucial in areas such as numerical linear algebra, where the stability of computed invariant subspaces is a primary concern, and in control theory, where system stability is analyzed through operator properties.