/mcpData sheet Cisco public © 2025 Cisco and/or its affiliates. All rights reserved. Performance numbers for all switch models C9500-16X C9500-32C C9500- 32QC C9500- 48Y4C C9500- 24Y4C C9500X- 28C8D C9500X- 60L4D ASIC UADP 2.0 UADP 3.0 Q200 Per-VLAN Spanning Tree (PVST) instances 3003 40,001 4,094 Spanning Tree Protocol (STP) virtual ports (port* VLANs) for PVST 13,000 16,000 16,000 STP virtual ports (port VLANs) for Multiple Spanning Tree (MST) 13,000 1,000,004 32,000 Total switched virtual interfaces (SVIs) 1,000 4,0001 3800 Jumbo frames 9,198 bytes 9,216 bytes 9,216 bytes 1 Varies based on selected flexible ASIC template and platform. 2 Line rate for 187-byte packet size and above. 3 300 with IOS XE Release 17.1.1 or later. 256 with IOS XE 16.12.x and 16.11.x 128 with IOS XE 16.10.x or earlier. 4 100,000 with IOS XE Release 17.5.1 or later. With IOS XE 17.4.1 or earlier: 32,000 with C9500-32C and C9500-32QC; 52,000 with C9500-48Y4C; 28,000 with C9500-24Y4C. 5 Flexible NetFlow requires IOS XE Release 17.8.1 or later. 6 Table maximum. The exact percentage of allocation will depend on specific IP/mask combinations. 7 Table maximum. Shared hardware resource. Maximum ACEs depends on number and type of ACLs configured across various interface types. Important notes Directly connected (or host) IP routes mean any /32 or /128 routes, including those that are learned indirectly (clients attached to switch’s own VLAN/SVI and those /32 prefixes learned over any routing protocols, such as over OSPF). Indirectly connected (or advertised) IP routes are subnet routes with a prefix other than /32 or 128 (for example: /8, /16, /24, etc.). UADP 2.0 based C9500-16X supports 32,000 adjacencies in hardware. So, they can support up to ~32,000 directly attached clients (including all adjacencies) in their own VLAN/SVI. UADP 3.0 based C9500-32C, 32QC, 24Y4C, and 48Y4C support 80,000 adjacencies for SVI, with Software Database Manager (SDM) template of distribution and 90,000 direct routes for all supported templates when a Layer 3 routed port is used. In both UADP 3.0 and Silicon One Q200 models, the exact hardware programming (hash efficiency) of IP/mask entries depends on the specific number and type of IP/mask prefixes.