"BUT same time when run show processes cpu history this command i find cpu utilization was upto 90%
Am i reading it correct or wrong or how to read these graphs"
Actually, it appears in the last 72 hours your CPU peaked at 100%, but doing what, we don't know.
Those graphs show both peak (instantaneous) CPU utilization and average CPU utilization.
As to "slow" traffic, on a switch, CPU utilization doesn't necessarily correlate with data throughput, whether high or low.
For CPU, besides how "busy" it is, we want to know how busy it was doing what (like information from your show processes CPU).
BTW, when using the show processes CPU, often it's useful to use the "sorted" option, if provided, and/or exclude lines with "0.00% 0.00% 0.00%".
For your stats, that would be something like:
4284 36090 2597758 0 8.20% 8.88% 8.88% - bcm_usd
4348 4248860 1969935 2 0.09% 1.42% 1.47% - statsclient
4280 3879030 11969037 0 0.38% 1.08% 1.17% - carmelusd
4112 3604950 881654 4 1.06% 1.14% 1.12% - pfma
4351 1787850 6115676 0 0.09% 0.29% 0.62% - snmpd
4251 580 459156 0 0.48% 0.47% 0.47% - sysinfo
4595 1462070 223311 6 0.00% 0.34% 0.43% - afm
4637 964540 397653 2 0.38% 0.30% 0.29% - udld
4630 693940 1574287 0 0.00% 0.08% 0.22% - ethpm
4422 370 9953749 0 0.19% 0.18% 0.20% - pktmgr
4419 300 2938175 0 0.09% 0.17% 0.17% - arp
4459 430 7470785 0 0.09% 0.11% 0.17% - netstack
4188 76330 23373 3 0.00% 0.21% 0.15% - fs-daemon
4344 257060 868182 0 0.00% 0.07% 0.08% - fwm
4627 154060 271318 0 0.00% 0.01% 0.04% - vlan_mgr
4598 73100 1131542 0 0.00% 0.04% 0.03% - stp
26596 1070 3318 0 0.00% 0.04% 0.03% 0 vsh
4255 82540 21354 3 0.00% 0.02% 0.02% - sensor
4364 180 12768758 0 0.00% 0.02% 0.02% - adjmgr
4519 91300 1680029 0 0.00% 0.02% 0.02% - cfs
4520 160 16100696 0 0.09% 0.02% 0.02% - ecp
4276 7820 59013 0 0.00% 0.01% 0.01% - ethpc
4421 160 90167 0 0.00% 0.01% 0.01% - icmpv6
26592 550 6574 0 0.00% 0.01% 0.01% - dcos_sshd
BTW, totaling those columns roughly matches the summary totals. Also, for those processes, nothing obvious, I believe, that would dramatically slow throughput.
Many things can impact data transfer rates beyond switch CPU utilization. Insufficient information to identify cause of server user performance complaints, or to even verify those complaints are valid.
