cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1857
Views
15
Helpful
8
Replies

Uplink interface speed between Core Catalyst9300 stack to Access Catalyst 9200 switch

4pratik
Level 1
Level 1

Hi Cisco Engineers,

 

I have question regrading interface requirement to create link between 2 catalyst stacks.

 

Switch 1 is 9300-Stack of 2 switch acting as Core Switch ( L3 ) 

Switch 2 is 9200-Stack of 3-4 switch configured as Access Layer switch (150 users + Accesspoints + Phones) 

 

I've port-channel between them using 2X1G links for now, can you please tell me if I need to increase interface capacity or 2G uplinks are fine?  Some designs I've seen has 2X10G links so wanted to make sure I need to do the same or not.

 

Any suggestions will be appreciated.

 

Pratik Patel

3 Accepted Solutions

Accepted Solutions

Hello

Look for high interface utlization for both egress/ingrees traffic, divide the number it shows by it max value and then multiply by 100 this gives you the % of traffic
Example:
show interface x/x | in tx|rx

txload 254/255 = 99%

rxload 100/255 = 39%

 


Please rate and mark as an accepted solution if you have found any of the information provided useful.
This then could assist others on these forums to find a valuable answer and broadens the community’s global network.

Kind Regards
Paul

View solution in original post

Joseph W. Doherty
Hall of Fame
Hall of Fame
"I've port-channel between them using 2X1G links for now, can you please tell me if I need to increase interface capacity or 2G uplinks are fine?"

Cannot answer that, with the information provided.

"I'll monitor portchannel , should I look for anything specific besides output drops on portchannel??"

Yes, many things. Perhaps the most important, and so far unmentioned, is how are the user network applications performing? For instance, you mention "phones". VoIP phones have demanding network service requirements. Many of which can only be met by providing sufficient bandwidth.

So, for example, if all 150 (?) VoIP phone were active, all using g.711, they would need about (100 Kbps * 150 phones) 150 Mbps, easily handled by a single gig link, assuming you're using QoS to insure the VoIP phone both obtain this bandwidth and, further, have priority over other traffic (to meet latency and jitter requirements).

Yet, since 150 gig connected users, could in theory, utilize up to 150 gig, even dual 10g uplinks could be insufficient. Only by monitoring network application performance, can you really determine your bandwidth requirements.

Something, though, to keep in mind. Many network applications are "elastic" in their bandwidth requirements. What they might want to use, or could use, can differ greatly with what they actually require.

As in the case of VoIP, with QoS you can manage bandwidth, insuring applications might obtain bandwidth they really need, at the expense of those with more "elastic" requirements.

Lastly, low link utilization doesn't always mean you have sufficient bandwidth, and conversely, high/full utilization doesn't mean you have insufficient bandwidth. Again, it really depends on network application (for acceptable performance) bandwidth needs. (BTW, my experience has been, using QoS, often allows you to use dramatically "less" bandwidth [because you use it much more efficiently].)

View solution in original post

The book answer is you should use QoS everywhere, end-to-end.

The real need for QoS is to apply only on interfaces which have congestion, which is actually adverse to the network applications, also assuming QoS can mitigate the adverse issue.

Of course, if you don't have QoS everywhere, you need to constantly monitor where it might be needed, but even when you do have it everywhere, you still need to monitor QoS, because sometimes QoS policies need revision to continue to be as effective and sometimes you do need more bandwidth.

View solution in original post

8 Replies 8

Leo Laohoo
Hall of Fame
Hall of Fame
Monitor the port channel traffic for about a month.
If the link gets saturated at 2 x 1 Gbps then upgrade to 2 x 10 Gbps.

Thanks Leo,

 

I'll monitor portchannel , should I look for anything specific besides output drops on portchannel??

 

Pratik Patel

Hello

Look for high interface utlization for both egress/ingrees traffic, divide the number it shows by it max value and then multiply by 100 this gives you the % of traffic
Example:
show interface x/x | in tx|rx

txload 254/255 = 99%

rxload 100/255 = 39%

 


Please rate and mark as an accepted solution if you have found any of the information provided useful.
This then could assist others on these forums to find a valuable answer and broadens the community’s global network.

Kind Regards
Paul

Thank you Paul,

Pratik Patel

Joseph W. Doherty
Hall of Fame
Hall of Fame
"I've port-channel between them using 2X1G links for now, can you please tell me if I need to increase interface capacity or 2G uplinks are fine?"

Cannot answer that, with the information provided.

"I'll monitor portchannel , should I look for anything specific besides output drops on portchannel??"

Yes, many things. Perhaps the most important, and so far unmentioned, is how are the user network applications performing? For instance, you mention "phones". VoIP phones have demanding network service requirements. Many of which can only be met by providing sufficient bandwidth.

So, for example, if all 150 (?) VoIP phone were active, all using g.711, they would need about (100 Kbps * 150 phones) 150 Mbps, easily handled by a single gig link, assuming you're using QoS to insure the VoIP phone both obtain this bandwidth and, further, have priority over other traffic (to meet latency and jitter requirements).

Yet, since 150 gig connected users, could in theory, utilize up to 150 gig, even dual 10g uplinks could be insufficient. Only by monitoring network application performance, can you really determine your bandwidth requirements.

Something, though, to keep in mind. Many network applications are "elastic" in their bandwidth requirements. What they might want to use, or could use, can differ greatly with what they actually require.

As in the case of VoIP, with QoS you can manage bandwidth, insuring applications might obtain bandwidth they really need, at the expense of those with more "elastic" requirements.

Lastly, low link utilization doesn't always mean you have sufficient bandwidth, and conversely, high/full utilization doesn't mean you have insufficient bandwidth. Again, it really depends on network application (for acceptable performance) bandwidth needs. (BTW, my experience has been, using QoS, often allows you to use dramatically "less" bandwidth [because you use it much more efficiently].)

Thank you Josh,


Pratik Patel

Hi Josh,

 

Thank you for valuable input, QQ: Should I use QoS on both switches or just core switch?

 

Pratik Patel

The book answer is you should use QoS everywhere, end-to-end.

The real need for QoS is to apply only on interfaces which have congestion, which is actually adverse to the network applications, also assuming QoS can mitigate the adverse issue.

Of course, if you don't have QoS everywhere, you need to constantly monitor where it might be needed, but even when you do have it everywhere, you still need to monitor QoS, because sometimes QoS policies need revision to continue to be as effective and sometimes you do need more bandwidth.
Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Review Cisco Networking products for a $25 gift card