z**r
发帖数: 17771 | 1
agree, but SP does NOT always want to provide classified services based on the
customers' COS. Different customers could have different COS for same kind of
requirement, say voice traffic. However, from SP's perspective, all the voice
traffic from different customers with different COS should be treated the
same. That's why SP might want to remark the EXP bits.
used
I thought 'set mpls exp imposition' setts both tunnel and vc labels, correct
me if I am wrong. sure vc label is not used for core |
|
b******e
发帖数: 66 | 2 The better question to ask would be:
1)cos is 0, why? 2) cos is 2, why? 3)cos is 4, why? |
|
z**r
发帖数: 17771 | 3 on the PEs I use mls qos trust cos, and mark the exp-bit for both tunnel label
and vc label, not just the tunnel lable. So on the egress PE side, after the
label disposition, the PE should copy the exp bits (in vc label) back to the
dot1q p bits, right?
the ce facing interface on the pe is just the cheap LAN card interface, which
is even not mpls awareness, but it DOES know mls qos.
If it marks only the topmost label, I would agree with you
QoS
are |
|
z**r
发帖数: 17771 | 4 multi services edge routers have different goals than PWE3 routers. MSRs are
more focused on terminating the different services, a lot of them are actually
broadband ras, and they have plenty of kinds of interface cards like you
described.
however PWE3 routers are more focused on transportation, you either deploy
VPLS/EoMPLS/VPWS, or other Martini stuff, or kempella (this is very new, cisco
might be slow on this).
I would say they both are routers, for an edge box, it really doesn't have to
run |
|
b******e
发帖数: 66 | 5
cisco
The reason Cisco did not implement Kempella draft for signaling is because:
Martini works for Cisco while Kempella (the older one) works for Juniper.;-)
Cisco will have Kempella implementation for VPLS siginalling and autodiscovery
pretty soon, I heard.
So overall, what do you think about the fate of small MSE vendors? will they
have a chance when competing with Cisco/Juniper/Alcatel in this space? what do
SPs really want? one MSE vendor in mind: www.hammerheadsystems.com, they did
severa |
|
z**r
发帖数: 17771 | 6 multi services edge routers have different goals than PWE3 routers. MSRs are
more focused on terminating the different services, a lot of them are actually
broadband ras, and they have plenty of kinds of interface cards like you
described.
however PWE3 routers are more focused on transportation, you either deploy
VPLS/EoMPLS/VPWS, or other Martini stuff, or kempella (this is very new, cisco
might be slow on this).
I would say they both are routers, for an edge box, it really doesn't have to
run |
|
z**r
发帖数: 17771 | 7 multi services edge routers have different goals than PWE3 routers. MSRs are
more focused on terminating the different services, a lot of them are actually
broadband ras, and they have plenty of kinds of interface cards like you
described.
however PWE3 routers are more focused on transportation, you either deploy
VPLS/EoMPLS/VPWS, or other Martini stuff, or kempella (this is very new, cisco
might be slow on this).
I would say they both are routers, for an edge box, it really doesn't have to
run |
|
z**r
发帖数: 17771 | 8 Look at this article, which mentions PBT by Nortel, what's that? any
differenece than VPLS/EoMPLS+EOAM? |
|
z**r
发帖数: 17771 | 9 主要是MPLS, VPLS和Carrier Ethernet，IPTV等wireline service方面的 |
|
z**r
发帖数: 17771 | 10 Cisco对新技术一向比较慢，尤其不是自家的技术的话。这个可以理解。就像最开始
vpls也是riverstone大搞特搞，cisco也不积极，后来还是很快adopt以后，成为领头军。
PBT基本上只有nortel支持，但是光PBT肯定不行，将来还是得靠PBB，而PBB应该马上就
会进入standard了，cisco没有理由不支持了就。 |
|
z**r
发帖数: 17771 | 11 说是他见过最stupid的一个技术。这个人是IEEE 802.1的大牛。不过还是很推崇PW |
|
T***y
发帖数: 458 | 12 文一个门外汉的问题，三年前在国内碰到一个台湾公司
非要发展串口网络技术(RS242 RS485)，组建工业现场网络，感觉是不是属于同样的愚
昧？
还是TCP/IP根本无法用在实时工业现场？
但是，很多老的概念好像都有轮回的一刻，就像串行通讯，某段时间呢大家都觉得不值得
发展，但是蛰伏了多年以后，这项技术概念有了突破，连硬盘都要SATA了
虽然技术跟以前完全不同了，但是概念还是从原始的演化而来的吧。 |
|
z**r
发帖数: 17771 | 13 你这个问题很尖锐其实。
技术之间相互借鉴太多了，x.25的技术都能应用到wireless上
值得 |
|
c****d
发帖数: 116 | 14 我觉得工业控制这块， 更多的是使用既有设备， 要求与现有的东西兼容， 另外
可能的原因是简单， 可靠。 工业现场环境通常非常恶劣， 干扰也非常大。
硬盘从PATA (parallel ATA) to SATA (Serial ATA), 我当初也觉得不可理解，
因为并行的输出理论上比串行的快的多。 我问过一个在Foxconn工作的人，他的
解释是当传输速率达到一定的上限后并行的传输干扰也非常严重， 效率也下降，
SATA的优势就显出来了。
值得 |
|
|
g**d
发帖数: 77 | 16 10G也有吧，我觉得metro e比较复杂，什么vpls，qinq，qos classification都在这个
layer上，而且变化也多，用npu比较好。到了core，就不需要了，反正是pkt fwd，pkt
的处理少了
我知道jnpr有现在的产品使用np2，ezchip的季报里是说，csco有些新产品，还在研制
中，08年量产。 |
|
|
z**r
发帖数: 17771 | 18 BRAS is used only for PPPoE based high speed Internet access and maybe VoIP
in I
PTV solution, video bypasses BRAS, but it's handled by EoMPLS/VPLS
architecture,
DSLAM can be a CE device |
|
|
z**r
发帖数: 17771 | 20 差不多，但是其实video的主要流量还是unicast，也就是VoD，也是通过EoMPLS/VPLS走
1483B |
|
c*****i
发帖数: 631 | 21 ios only support lsp signalling. xr will support bgp-vpls in next release. |
|
s*****g
发帖数: 1055 | 22 That is for L3VPN inter-AS, has nothing to do with L2VPN (VPLS)'s BGP
signalling. |
|
s*****g
发帖数: 1055 | 23 Lao Da, that is for VPLS neighbor auto-discovery, not signalling. |
|
d****i
发帖数: 1038 | 24 Mostly for business and political reasons, partly for belief and technical
reasons, :) IOS won't be allowed to support BGP-VPLS. but IOS-XR decided to
go ahead to support it to counter some of the Juniper efforts. :)
on |
|
c*****i
发帖数: 631 | 25 confused.. you are trying to use ldp signaling or bgp-vpls autodiscovery/
signaling? |
|
z**r
发帖数: 17771 | 26 对，eompls建立的vc和svi通过vlan或者bridge domain或者connect联起来
去。
interface, |
|
z**r
发帖数: 17771 | 27 另外如果通过vlan连接那么vc number和svi对应的vlan number要一致，如果BD活着
local connect就无所谓了。
不过你这个doc是关于OSM的，已经属于被淘汰的产品了，你最好看sip-400，es-20 or
ES+的document，里面会有一个革命性的EVC，感觉颠覆了很多传统vlan的一些东西 |
|
|
s*****g
发帖数: 1055 | 29 atom? I was majoring in atom/molecular physics back in college, I am an
expert in atom, shrodinger equation etc. Just kidding, although they all belong to MPLS L2 VPN category, Cisco seems to differentiate among EoMPLS, VPLS, and AToM which specifically refers to PPP/HDLC/ATM/FR over MPLS. |
|
d****i
发帖数: 1038 | 30 not really. actually, AToM in Cisco mainly means ldp signaled pseudowire.
belong to MPLS L2 VPN category, Cisco seems to differentiate among EoMPLS,
VPLS, and AToM which specifically refers to PPP/HDLC/ATM/FR over MPLS. |
|
t*******r
发帖数: 3271 | 31 唉这位老大非要说英文不可嘛~~~~看得我怪累的.
我倒是觉得如果是说MPLS VPN, 指的是二层三层都有的.
还有PPVPN只是隔离, 没有加密.
belong to
MPLS L2 VPN category, Cisco seems to differentiate among EoMPLS, VPLS, and
AToM
which specifically refers to PPP/HDLC/ATM/FR over MPLS. |
|
z**r
发帖数: 17771 | 32 AToM主要用来和基于L2TPv3等tunnel的L2VPN区分的吧？如果说MPLS L2VPN俺觉得就应
该是AToM
belong to MPLS L2 VPN category, Cisco seems to differentiate among EoMPLS,
VPLS, and AToM which specifically refers to PPP/HDLC/ATM/FR over MPLS. |
|
c*****i
发帖数: 631 | 33 l2vpn有基于mpls和l2tpv3的。然后又分vpws和vpls。signalling方面又有ldp和bgp。
然后又有bgp auto discover+ldp或者bgp singalling。
, |
|
s******v
发帖数: 4495 | 34 backhaul不是太明白，我看我理解的对不对
ce1 --- peA (as 1) -- ..inter-isp cloud.. --- peB (as 2) --- ce 2
1) ce1 = hub base station router, 有自己的vlan, 例如100, 还有多个lan，每个
对应一个leaf base station;
2) ce2 = att msc， 这边是什么，ip? 还是vlan trunk?
感觉inter as l2vpn vpls是可以的，不过要两个isp都support，有点复杂。
最好还是 mpls l3vpn, 每个基站给个/24 network, peA - ce1 and peB - ce2用
ebgp，中间是 inter as opt b/c，不过这个isp的事。
于距离
MSC),
site |
|
s*****g
发帖数: 1055 | 35 Why would any body care how a box internally works? and what is the
difference between a switch and routers nowadays?
Difference between 7600 WAN ethernet and LAN ethernet not just buffer and
QoS, internally packet forwarding path is different also. WAN ethernet was designed to have MPLS forwarding capability before
Sup720 came out, while LAN ethernet has to rely on PFC to switch MPLS
packets with PFC3BXL and newer versions of PFC, LAN ethernet can not be used
for VPLS uplink. All other AToM fe... 阅读全帖 |
|
a***n
发帖数: 262 | 36 审美疲劳 is one factor with us too :-)
cisco的asr9k应该是目前看最牛逼的ethernet router/switch. We just have Cisco
team proposing ASR9k to replace our campus 6500 backbone. Some features in
ASR9K like VPLS, EoMPLS that we would like to explore. But I am confused
that there is no zone based firewall, IOS firewall or FWSM for ASR9K. We
have ACLs and FWSMs for our 6500. Is ASR9K positioned to service provider
only? Or we just have to use external ASA for security? The other thing for
ASR9K, the initial backplane ... 阅读全帖 |
|
s*******8
发帖数: 12734 | 37 学校里应该最喜欢讲TCP/IP吧。
其实理解了非常重要，比如昨天遇到道题，VPLS需要多少pseudowire，我一条条排列的
数过来，最后发现就是相当于full mesh，哈哈。 |
|
t*******r
发帖数: 3271 | 38 来自主题: EmergingNetworking版 - 一个面试题 多台router做RR主要是: 1, 系统资源分开; 2, 便于管理;
PE上多个loopback是便于管理, 但系统资源基于业务的分配(IPv4/IPv6/L3VPN/VPLS)目
前厂家实现是没有
能真正分开的. |
|
s*****g
发帖数: 1055 | 39 One instance corresponds to one VC id, within the same VCid, circuit type
must match, if you are asking one peer, different VC type, then it is OK. |
|
s*****g
发帖数: 1055 | 40 RFC probably does not care about this situation, vendor should implement in
a way that if VC-type mis-matches, it will automatically adjust one side VC-
type in order to bring up the VC. |
|
d****i
发帖数: 1038 | 41 sure we can configure other PEs in a way that brings up VC, and we can
configure in the way one VC type 4,
one VC type 5 and both VCs are up. So it is not a violation to any RFC? It
will certainly add more complexity for
vendor's products though.
in
VC- |
|
t*******r
发帖数: 3271 | 42 我觉得自适应会带来非技术性的其他问题.
自适应? 谁适应谁? 互不相让, 这不就打起来了嘛~~~
in
VC- |
|
s*****g
发帖数: 1055 | 43 It is a very simple state machine. |
|
d****i
发帖数: 1038 | 44 basically, if one end is 4 the other end is 5, both end will become 4. and
once in 4, it cannot become 5 again
to prevent pingpong. |
|
z**r
发帖数: 17771 | 45 I don't see why it cannot, but it'll depend on the vendor implementations ... |
|
|
z**r
发帖数: 17771 | 47 negociate最起码的原则就是相互商量啊，得有个标准啊，就像OSPF/ISIS里谁做DR，谁
做BDR，STP里谁做root，10M/100M/1000M到底应该是多少，等等 |
|
b******s
发帖数: 5329 | 48 语音最近几年没搞了。sp环境单一，每天就是igp, bgp, mpls，vpls, label,tunnel,
再不就是76, asr, crs,linecard, RSP, MX，T。。玩段时间就累了。 |
|
s******k
发帖数: 234 | 49 来自主题: EmergingNetworking版 - 坑爹的思科培训 思科公司的培训最废柴。我思科的培训大大小小几十近百个，最开始是各种认真，各种
笔记，各种复习找资料。现在就是各种睡，凑人数。全是讲概念，各种概念，各种潮流
，各种发展，各种市场广阔，各种未来增长。谁关注这个，我要得是实实在在的东西。
我现在唯一印象深点得就是一次马提尼亲讲的VPLS和有一次的nexus s的数据中心还算
有点货。。。。 |
|
z**r
发帖数: 17771 | 50 otv用unicast做signaling的时候，对site数目限制，用multicast的时候，估计也会有
限制，但是这个数目是多少呢？不过别忘了vpls一样也有限制。
之所以想跟大家聊聊这个，主要最近看到有些需求是data center consolidation中，
有倾向是cross site layer 2 + ipv6 only。
另外，otv以前是不支持ipv6的，现在的支持怎么样了？
limitaton |
|
