Dipak wrote: > What about Efficient SpeadStream 3010 card? Does it support shaping? Yes, although it's essentially limited to one CBR vc per card (and as many UBR vcs as you want). There are some other things that the card can do but they're pretty esoteric. > For UBR, how much bandwidth will be allocated for the first VCC opened. "U" stands for unspecified. The card will do best-effort at taking all the cells it can. > For > subsequent VCCs how the bandwidth will be distributed? Any UBR VCC can use all the available bandwidth. If more than one are going full blast, hopefully the hardware will do a decent job of round-robining them. > How one can > support VBR? Depends on the hardware, I'd suppose. > > There's also a ->change_qos() entry point which can be called to [...] > What you will be going to do in this function? The function is allowed to change the rates on the card. So if your hardware can open a CBR vcc and then change the bandwidth dynamically, you could provide this function. > Even if you change the > QOS of your VC at your end, changing at the network of an established > VCC is not possible, I suppose. So, how much one will be gaining from > it? For SVCs the singalling code will ask the network for a new QoS before calling your card's change_qos() function. But, yes, from the perspective of the driver you only are worrying about what shaping your card can do. -Mitch - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.rutgers.edu
