[c-nsp] ME3600X buffer calculations

[Waris Sagheer (waris)]

Thanks Darren.
Me3600X has 44 MB of buffer. There are two ASICs in the system. Each ASIC has 22 MB of buffer.24x1Gig are part of one ASIC and 2x10Gig ASIC are part of the second ASIC. In case if you are not using 10Gig SFP+, you can use 1Gig SPF in the 10Gig slot.
Out of 22MB, 4 MB are used internally so usable buffer size would be 18 MB per ASIC. Total buffer per system would be 36 MB. Buffers are at the global level so resources are not assigned per interface. There is an intelligent resource allocator within the system which assigns resources to interfaces as per request.
512 PW are supported on ME3600X so you can use all of them with any feature combination. Whenever a queue is configured, default q limit is applied. Below are the default Q limit values. In 15.3(2)S, default Qlimit for Gig interface would be 48KB and maximum Qlimit is 2MB. The software allows oversubscription so you can configure higher Q limit values on all the queues since we do not expect all the queues to be congested at the same time. In case if you need higher limit value than 2MB, you can use the flexible Qlimit feature which allows Qlimit to be configured in percentage. You can assign 100% of buffer. In summary, you can configure all the queues with the desired Qlimit without any issues.



Default Q-limit (Prior Release 3.7 Release)


In Bytes:

10/100/1000/10000 Mbps - 12/12/12/120 KBytes respectively

In Time:

10/100/1000/10000 Mbps - 10000/1000/100/100 usec respectively


Queue-Limit Range & Maximum Queue-Limit


200 to 491520 Bytes

1 to 3932 us

1 to 2457 packets (Assuming 1 packet = 200 bytes)

491520 Bytes (Maximum Queue Limit)

*CLI Range is shown higher than the platform can support. There is an issue with the command help which shows higher value than supported by the platform,


Default Q-limit (Release 15.2(4)S Release)


In Bytes:

10/100/1000/10000 Mbps - 12/12/48/120 KBytes respectively

In Time:

10/100/1000/10000 Mbps - 10000/1000/400/100 usec respectively


Maximum Queue-Limit (Release 15.2(4)S)


2 MBytes


Percentage Based Q-Limit (Release 15.3(1)S)


Q-limit can be configured up to Maximum Buffer size 44MB for ME3600X/ME3600X-24CX & 352MB for ME3800X


Best Regards,

[http://www.cisco.com/web/europe/images/email/signature/horizontal06.jpg]

Waris Sagheer
Technical Marketing Manager
Service Provider Access Group
waris at cisco.com<mailto:waris at cisco.com>
Phone: +1 408 853 6682
Mobile: +1 408 835 1389

CCIE - 19901


<http://www.cisco.com/>



[Think before you print.] Think before you print.

This email may contain confidential and privileged material for the sole use of the intended recipient. Any review, use, distribution or disclosure by others is strictly prohibited. If you are not the intended recipient (or authorized to receive for the recipient), please contact the sender by reply email and delete all copies of this message.

For corporate legal information go to:
http://www.cisco.com/web/about/doing_business/legal/cri/index.html



From: Darren O'Connor <darrenoc at outlook.com<mailto:darrenoc at outlook.com>>
Date: Saturday, July 27, 2013 3:22 PM
To: Waris Sagheer <waris at cisco.com<mailto:waris at cisco.com>>
Cc: "cisco-nsp at puck.nether.net<mailto:cisco-nsp at puck.nether.net>" <cisco-nsp at puck.nether.net<mailto:cisco-nsp at puck.nether.net>>
Subject: Re: [c-nsp] ME3600X buffer calculations

Hi Waris.

15.3(2)S

Mainly as reading through previous threads it was suggested to use this as it now has the queue-limit available

Thanks

Darren

http://www.mellowd.co.uk/ccie

On 27 Jul 2013, at 21:30, "Waris Sagheer (waris)" <waris at cisco.com<mailto:waris at cisco.com>> wrote:

Darren,
Which image you are using? Based on that I can send you the write up.

Regards,
-Waris

On Jul 27, 2013, at 8:27 AM, "Darren O'Connor" <darrenoc at outlook.com<mailto:darrenoc at outlook.com>> wrote:

Hi all.

I've seen that the ME3600X has 44MB of buffer space for packets. I'm trying to work out how many virtual circuits I can push through the box before my buffers run empty.

I know this is a difficult question, as a lot depends on how much traffic is going through each EVC at any one time. I need to know how to work it out though.

As an example, let's say I'm using 12 ports each at a gig. Each port carries multiple point to point links separated by vlan tag. Certain customers would be paying for QoS and hence I'll shape to the needed speed and give priority to certain frames in a child class. If a customer were not paying for QoS I could simply police that customers frames going through the ME3600X.

What maximum figures am I looking at in regards to maximum shaped traffic at any particular time? Or at least, how can I work this out?

Thanks

Darren

_______________________________________________
cisco-nsp mailing list  cisco-nsp at puck.nether.net<mailto:cisco-nsp at puck.nether.net>
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/