cancel
Showing results for
Did you mean:
cancel
Announcements

209
Views
15
7
Replies
Beginner

## Nexus input output rate

I am getting confused with the input output rate of an interface, how can my input rate be less than my output rate?  example below:

30 seconds input rate 11246640 bits/sec, 20009 packets/sec
30 seconds output rate 1650525584 bits/sec, 136182 packets/sec
input rate 11.25 Mbps, 20.01 Kp; output rate 1.65 Gb 136.18 Kb
Load-Interval #2: 5 minute (300 seconds)
300 seconds input rate 11894688 bits/sec, 21154 packets/sec
300 seconds output rate 1785522904 bits/sec, 147285 packets/sec
input rate 11.89 Mbps, 21.15 Kpps; output rate 1.78 Gbps, 147.29 Kpps

7 REPLIES 7
VIP Expert

## Re: Nexus input output rate

". . . how can my input rate be less than my output rate?"

Because, generally, the same traffic flows don't ingress/egress on the same interface, More common, ingress flows are on one interface and the corresponding egress is on a different interface. Further, ingress flows one interface might go to different egress interfaces and conversely egress flows might be coming from several ingress interfaces.

What usually balances out is the total/aggregate of all interfaces ingress rate with the total/aggregate of all interfaces egress rate. However, the later might not be true if interface ACLs, or policies, are dropping some traffic, or an egress interface is congested enough that it's dropping packets.
Hall of Fame Master

## Re: Nexus input output rate

Joseph presents a good explanation based on the possibility that data ingress is one interface and data egress is on a different interface. I would suggest another explanation which can also be considered. Think about the possibility that a packet inbound is from a client and is a request to a data base server for information. The request was 1 packet. The response from the data base server has lots of data and it takes 10 packets to transmit it all. So at that point we have a 10 to 1 difference in the output to input rates.

HTH

Rick

By doing so, and until end of January, you are helping Doctors Without Borders
VIP Expert

## Re: Nexus input output rate

Rick makes an excellent point! What he is describing is traffic volume, between hosts might (and often is) asymmetric. However, that would not change or impact what I described in my first post.
Hall of Fame Master

## Re: Nexus input output rate

Joseph

Certainly my explanation does not impact your explanation. Both are valid explanations of why the input rate may be less than the output rate for an interface. Perhaps we understand some terms differently. I would think that the term asymmetric would apply more to the explanation of input on one interface and output on a different interface. My explanation is perhaps more about balance, where one input received may generate many output responses sent. But certainly both are valid explanations of the original question about whether it is legitimate to have input rate less than output rate for an interface.

HTH

Rick

By doing so, and until end of January, you are helping Doctors Without Borders
Highlighted
VIP Expert

## Re: Nexus input output rate

Rick you're spot on in how various people can understand differently (terms, of course, too).

I whole heartily agree, your explanation doesn't impact mine (but by writing that, does it seem to imply I believed your explanation did?).

I was concerned that less experienced readers might not realize that our two explanations are not exclusive. I.e. as you say both are valid, but again, they aren't exclusive of each other. It was that which I was trying to make clear. Perhaps, as I often I do, I only muddied the waters further by how I tried to express that.
Hall of Fame Master

## Re: Nexus input output rate

Joseph

I interpreted your comment as attempting to address the possibility that some readers might think that one of our explanations could impact the other. Both are quite valid and they are independent of each other. As you and I have learned there are frequently more than one way to solve a network problem, and more than one way to explain a network behavior. And I believe that if we thought about it some more we could probably come up with another reason or two for this behavior.

HTH

Rick