According to Menger’s theorem, connectivity and edge connectivity are closely related to node-disjoint paths and edge-disjoint paths, respectively. Node- and edge-disjoint paths can keep the effective transmission of information and confidentiality. Therefore, node-disjoint paths and edge-disjoint paths are two important parameters to measure the reliability of a network. For a faulty node (resp. edge) set $S\subset V$ (resp. $S\subset E$), a connected graph $G=(V,E)$ is $S$-strongly Menger-node-connected (resp. Menger-edge-connected) if any two distinct nodes $x$ and $y$ in $G-S$ are connected by $\min \{\deg _{G-S}(x),\deg _{G-S}(y)\}$ internally node-disjoint (resp. edge-disjoint) paths in $G-S$, where $\deg _{G-S}(x)$ and $\deg _{G-S}(y)$ are the degrees of $x$ and $y$ in $G-S$, respectively. And most of the previous studies are based on networks that are triangle-free. In this paper, we consider the strongly Menger (edge) connectedness of a class of $r$-dimensional recursive networks RNCG $G_{r}$ with triangles. Moreover, we show that $G_{r}$ is $(rl-l-1)$-strongly Menger-node-connected. And then we show that $G_{r}$ is $[k+(r-1)l-2]$-strongly Menger-edge-connected of order 1 and $[2k+2(r-1)l-6]$-strongly Menger-edge-connected of order 2. Since the class of $r$-dimensional recursive networks RNCG $G_{r}$ includes not only data center networks DCell and generalized DCell but also interconnection network dragonfly, etc., all the results are appropriate for these networks.