Store & Forward vs cut-through switching

[ajay chauhan]

Hey Guys,

Looking for expert opinion on what layer (AGG/Tor) what type of packet forwarding should be used specially in DC.

is there no buffering at all when we use cut-through ? I understand on TOR Store and Forward only make sense if high buffer cheap switches are being used.

On cisco.com i found for nexus 3k -• Store-and-forward mode activates automatically for a port when the switch identifies that the port is
oversubscribed and the ingress rate is greater than the switching capacity of the egress port.For example,
when the port ingress rate is 10 gigabit and the switching capacity of the egress port is 1 gigabit

However on other hand for 5 K-  10 G to 1 G is Cut-through.

Forwarding Mode Behavior (Cut-Through or Store and Forward)

SOURCE INTERFACE	DESTINATION INTERFACE	SWITCHING MODE
10 GigabitEthernet	10 GigabitEthernet	Cut-Through
10 GigabitEthernet	1 GigabitEthernet	Cut-Through
1 GigabitEthernet	1 GigabitEthernet	Store-and-Forward
1 GigabitEthernet	10 GigabitEthernet	Store-and-Forward

This makes me little bit confused appreciate if someone can clarify.

Thanks

Ajay

[Joseph W. Doherty]

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

Sorry, I'm confused about what you're confused about.  If you're asking about 3K Nexus forwarding operation, and why it is what it is, that's something only Cisco can answer.

If you're asking about store-and-forward vs. cut-through, the latter is designed to decrease store-and-foward latency.

Historically, when 10 Mbps switches first came out, they added latency (remember, its per hop) not seen with 10 Mbps hubs. When 100 Mbps came out, store-and-forward latency decreased so the need for cut-through fell out of favor.  (BTW, keep in mind there are issues starting forwarding a frame before you know you have a good copy.)

With 1g, 10g, 40g and 100g, store-and-forward latency is decreased even more but now we have applications that required ultra low latency.  We also now perhaps have applications using jumbo Ethernet.  So, there's been a bit of a revival of cut-through.

PS:

BTW, cut-through cannot work with ingress having a slower bit rate than egress.

[a12288]

The philosophy behind using cut-through in DC switch is that, you should have clean and clear physical layer and short distance of network connection in DC, and DC switches will add no benefits of doing CRC checks by using store-forward, because there should be none in DC. Instead, DC switch is to forward traffics as fast as possible, and let upper layer to handle the packet corruptions, if any.