Disclaimer
The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.
Liability Disclaimer
In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.
Posting
Sizing a (software) router really depends on amount of traffic that will transit it, packet sizes and what needs to be done per packet. I've attached a Cisco document that discusses many aspects of router performance referencing the later ISRs.
Often (software) routers can't handle LAN type interfaces at full line rate, but this doesn't mean they can not be used on LANs.
Assuming a (software) router can only handle 25% of line rate, what that means is as long as the average utilization doesn't exceed that 25% it will, on average, have enough time to process each packet. So, if average line utilization doesn't exceed 25%, your router can "keep up".
If more traffic is sent to the router than it can process, it will start to drop packets.
In ye olde days, performance specs were often only provided for maximum size packets, such requiring the least per packet processing, and thereby, allowing the greatest throughput numbers (which looked good on sales spec sheets). I.e., actual performance with typical traffic was worst.
Now a days, vendors often provide the spec for minimum sized packets, so typically you will obtain higher throughput, 3 or 4 times not uncommon. However, you need to account for duplex traffic and to try to account for extra service overhead, per packet, beyond just forwarding a packet. You also need to allow some CPU for other router services, not related directly to forwarding packets, e.g. routing protocols, SNMP, syslogging, etc.
I've found, if you take the overall PPS rate, translate it into minimum size bps, half it (for duplex), this sets (about) the upper boundary for "typical" packets bandwidth consumption. Halving this again, sets another boundary which allows for traffic growth and traffic being more than expected. If expected traffic is less than the lower boundary, probably a smaller router would be fine.
Trying to size a router by number of users is only good if you can predict the amount of bandwidth each user is going to use. There can be a case where something like an 880 is inadequate for just one user, yet in another case it might well support 200 users.
Yes an 880 with five FastE ports has the potential to be easily swamped, but there's often a huge difference between potential and actual usage.
