08-17-2011 01:38 PM
Enhanced Group Management (EGM) is a FOS-enabled license to ensure that
Brocade devices can be managed out of the box by DCFM. Specifically, EGM
enables simultaneous management of groups of switches and devices in terms of
configurations, firmware downloads, and more.
Is a optional License for Extended ( Long Distance ) Connectivity
08-17-2011 01:45 PM
Extended fabric helps on ISL and trunks. You can increase the number of buffers for each link. This can increase the number of frames which can be transported on that link. If you have links longer than 10km it is very important.
In case of shorter links is can also helps in a limit way.
Increase performance means not more MB per seconds it means more IOs with many short frames.
Enhanced Group Management is required for DCFM.
Adaptive networking give you more flexibility in traffic handling. Like Ingress Rate Limiting QoS Bottlenek detection.
For more information please consult the Admin Guide which gives a good introduction.
I hope this helps,
08-17-2011 04:46 PM
" it means more IOs with many short frames".. how it is significant here? I assume the packets get transffered in the MTU size, by deafult i am with 1500. So how does buffer helps in increasing the number of frames? how does that work?
08-18-2011 08:34 AM
a MTU of 1500 is a ethernet topic and belongs not to FC world.
A FC fame is up to 2148 bytes long. The payload can be up to 2112 bytes. If an ISL is established both switches take care about the number of buffers the partner can handle. One buffer is given away means one frame is on the fly. The sending switch gets it buffer back if the frame was transmittet. The frames can have a very short lenght or up to 2112 bytes. If you have many short frames you will see less throughput on the link. But the switch is not able to send more. From the throughtput you would say the switch can carry more load but this is wrong. If the buffer credit Zero time is hight you may have an IO issue and not a bandwidth issue. But if you increase the number of buffers that you can send more small IOs over the link.
The target is to fill the link with data. You need to know your average frame size. Then you can calculate the frame length and if you divide the link lenght by frame lenght you know the amount of frames you can put on the link. The number of frames is eq to the needed buffers.
And by the way If you change and ISL to LE mode the buffer handling will be changed by the switch. With LD or LS mode you can assign more buffers to a port. This is only possible if Extended Fabric Lic is present.
I hope this make sense,
08-18-2011 09:28 AM
Nice explanation. So this "buffer" is much close to what i think "buffer credit" . In my knowledge , Buffer credit is adjusted automatically by the switch.Is this correct and if yes in which mode buffer credit is adjusted automatically (ISL or LE)?
08-18-2011 12:55 PM
Yes buffer and buffer credit is the same.
Buffer Credit information will be exchanged between switches during link init phase.
Brocade currently support links with a max distance of 10km without Extended Farbic on full speed.
Which basicly mean that a fixed number of buffers are assigned to that link (ISL). There is no adjustment in terms of optimasation if you have smaller frames the switch will not add automaticly more buffers. It is justed a fixed number of buffers which can run a link upto 10km.
With Exended Fabric you can increase the number of buffers to the ISLs. You need the license on both ends. This can improve the number of IOs on that ISLs and increase the perfromance of applications as well. If you have lets say a 20km Link at 8 Gbit and no Extended Fabric lic --> the link will perform as an 4Gbit link in a best case scenario.
LD mode will mesure once during the link init process the latency of the link and calculates the needed buffers. After that this number of buffers do not change until the link get an link reset.
The good think is if you have a very slow TDM card in your connection the switch will see the total latency of your connection not only the fiber length. So the extra demand for the TDM cards will be included.
I hope this helps,
08-20-2011 02:37 AM
Chapter 20, Managing Long Distance Fabrics in Fabric OS Administrator's Guide (53-1001763-01) offers a good reading.
Another good option is SAN Distance Extension Reference Guide, also from Brocade (GA-SG-061-00):
Hope this helps,
10-04-2013 05:11 AM
im sorry to hijack this thread, but i am looking for help with interconnecting two datacenters 30km apart using 4Gbps ELWL 30km 1310nm SFP's (i only have one pair for now)
i have a long haul dark fibre connection between the 2 sites and from the dark Fibre patch panel i run 16m LC/UPC SM cable to my B80.
i have set the portcfglongdistance 4 LS 1 30 but still "no_light" status on the link/ports.
what have i missed? surely if the SFP transmit strong enough signal the light shoudl travel all the way but im worried that maybe too much db loss along the way because my Fibre Planner said our dark fibre is optimised for 1510nm wavelength and i might loose at least 40% at 1310nm wavelength. is this true for brocade SFP's also?could this be my problem?