d****i
发帖数: 1038 | 1 对， 要求就是没有IGP的情况下mpls还能跑起来，但是哪个哥们在配置的时候搞错一个
小地方就得查半边才能搞明白。
我觉得有点象钓鱼工程啊， 现在说得这也好那也好，真用起来了需要更新这个更新哪
个， 最后成本不一定低。哈哈
对了，7600马上要支持TP了。 7600， 6500这些老同志还真的不能退下来啊。 |
|
W****2
发帖数: 297 | 2 MPLS-TP is a good player in mobile backhaul. |
|
s*****g
发帖数: 1055 | 3 I feel the first half of the book (talking about MPLS infrastructure) is
good, but the second half (talking about applications) is just so so,
authors talk about a simple idea over and over again. |
|
|
d******t
发帖数: 834 | 5 PE-PE iBGP
PE-P-PE OSPF (接口开启MPLS)
PE-CE RIP V2
两个CE上分别用loopback口模拟192.168.1.0和10.1.1.0
路由表一切正常，两个CE可以看到对方的私有网段，ping不通，啥原因？ |
|
x*********n
发帖数: 28013 | 6 PE-pE里面iBGP，如果iBGP里面有个link变化了，那么对整个MPLS有什么影响？
packet还按照LIB传么？router会drop这个packet么？
欢迎大家指点迷津。 |
|
t*******r
发帖数: 3271 | 7 link就是link, 什么叫IBGP的link?
link断了, RSVP LSP重建, 重新分配mpls label.
业务断一小会儿, 没什么的.
frr开了的话, 断更小一会儿. |
|
b******s
发帖数: 5329 | 8 mpls vpn hub & spoke做法就是“脱了裤子放屁”，迎合某些特殊要求或似懂非懂的客
户，其实没啥实用价值。 |
|
|
z**r
发帖数: 17771 | 9 for your 1st question: cisco routers would rely on CEF table (adj table).
for your second question: come on, just read
ios>
else
still |
|
l***y
发帖数: 791 | 10 If in core network running a mix of the two mode (i.e. multi-vendor
environment) would it be any worse than running just just one? Junos only
supports ordered control but i think cisco runs indepedent.
my thinking is that at such time when new link comes up, you'd drop packets
for a few seconds regardless of which mode, just the location of the
blackholing would be different.
it seems that igp sync is good for these occassions, but I am guessing it
must require ordered control, and also depends ... 阅读全帖 |
|
s*****g
发帖数: 1055 | 11 There is no difference operation wise between ordered control vs.
independent control, just one extra configuration command, both modes are
equivalent. LDP has no idea and does not care what mode its neighbor is
running. Having said that, there is really not much practical use to have
any non/32 FECs. Only thing BGP/IGP shortcuts, VPN transport LSP need is a /
32 FEC
IGP/LDP synchronization has nothing to do with ordered or
independent control or size of LFIB. I believe IOS by default enables IG... 阅读全帖 |
|
s*****g
发帖数: 1055 | 12 Right LIB!=LFIB, but one is the other one's (exact) copy and they are always
in sync. In theory, you can not predict how many LIB entries you will ever
get from your downstreams, and you don't know how much process power your
neighbor has, IGP/LDP synchronization is designed to coordinate between to
LDP peers and IGP to avoid MPLS forwarding blackholing regardless label-
mapping size or control plane processing power. Does my reasoning make
logical sense?
When you say "I think", you'd better giv... 阅读全帖 |
|
K****s
发帖数: 59 | 13 问一个MPLS-VPN的问题：
需求是： 终端的CE需要跟远程data center通信，但是CE之间不可以通信。
可以想到的解法是data center可以跟每个ce之间用唯一的RD标识一个VPN连接，在CE端
可以用RT来控制接受数据，但是如果CE数目多的话，DATA CENTER的VRF数目也会增加。
请问什么解法效率最高？ |
|
s*****g
发帖数: 1055 | 14 Yes, but if PE1 has two or more CEs attached, then the two CEs will be able
to communicate with each other without going to hub unless you use different
VRF, which is not scalable. You need "scalable MPLS-VPN hub-spoke" feature
to circumvent this limitation. |
|
m**k
发帖数: 290 | 15 for control signaling.
可是 Juniper PTX 是 MPLS only. 这个怎么work? |
|
|
s*****g
发帖数: 1055 | 17 In deed MPLS does not need IP once LSP is up, but what about signaling? |
|
x*********n
发帖数: 28013 | 18 我太弱了，等我整理好BGP notes，做完ospf，再好好看看mpls |
|
a**********k
发帖数: 1953 | 19 When G-MPLS is used in Sonet/DWDM network, in theory it can do signalling on
any transport protocols(PPP, or HDLC?), but most
of the times, it will still be on some type of out-of-band or
in-band IP stack. |
|
|
z**r
发帖数: 17771 | 21 俺不这样认为。
事实上，不同的环境中并行着好几种。
拿mpls来说，单纯的mpls并不能给用户多大的好处，但是，mpls是几乎所有从edge到
backbone的基础。90年代末开始，各家公司就开始推，当时的主导方向差不多就是mpls
vpn和mpls te，有些用户，比如sprint，就认为mpls并不能给他带来什么实际的好处，但
后来慢慢认识到mpls vpn的好处，可是还是不想把自己的pure ip backbone升级到mpls
backbone，于是，又出现了mpls vpn over ip tunnels等等所谓rfc2547bis。基本上人们
都接受了mpls vpn这个了，于是什么carrier supporting carrier，mpls vpn awareness
IPsec等一堆东西也随之出现。
这个可以说mpls的第一波hype，现在，基本上处于第二波，也就是AToM，带宽越来越便宜
了，人们已经不满足相对复杂的layer3 backbone了（对于大型enterprise来说），很多
人希望有简单的layer2 backbone，于是人们最喜欢的传输方式,e |
|
z**r
发帖数: 17771 | 22 俺不这样认为。
事实上，不同的环境中并行着好几种。
拿mpls来说，单纯的mpls并不能给用户多大的好处，但是，mpls是几乎所有从edge到
backbone的基础。90年代末开始，各家公司就开始推，当时的主导方向差不多就是mpls
vpn和mpls te，有些用户，比如sprint，就认为mpls并不能给他带来什么实际的好处，但
后来慢慢认识到mpls vpn的好处，可是还是不想把自己的pure ip backbone升级到mpls
backbone，于是，又出现了mpls vpn over ip tunnels等等所谓rfc2547bis。基本上人们
都接受了mpls vpn这个了，于是什么carrier supporting carrier，mpls vpn awareness
IPsec等一堆东西也随之出现。
这个可以说mpls的第一波hype，现在，基本上处于第二波，也就是AToM，带宽越来越便宜
了，人们已经不满足相对复杂的layer3 backbone了（对于大型enterprise来说），很多
人希望有简单的layer2 backbone，于是人们最喜欢的传输方式,e |
|
w***s
发帖数: 321 | 23 其实说MPLS VPN也误导了，就是标准的MPLS + BGP结构，一样的BGP free Core效果，
只是用IPinIP模式，MPLS就可以省了。但是在MPLS方案里面PE还是有全路由，所以LISP
最重要的就是Map－Request功能。
之所以说这个更象H323查询过程，而不是DNS，原因有两个：
1. 系统里面已经有DNS，再搞一个会混淆
2. DNS没有做EID-RLOC转换，还是EID的不同表达方式，而从电话号码到Gateway的映射
更类似于EID-RLOC的关系，Admission Control就不是重点了，不过加上这功能也不错
，实际上在MAP Server上是可以实现一定级别的路由策略的。
最后的问题就是Tunnel End Router选取，取PE，就恢复到原始的MPLS＋BGP方案，MAP
Cache的规模多少近似于全路由。CE会好的多，毕竟一个site的用户不会访问整个
Internet，当然DC除外。如果允许有Default RLOC的话，整个方案看起来很不错。
running MPLS, service provider can still run... 阅读全帖 |
|
w***s
发帖数: 321 | 24 其实说MPLS VPN也误导了，就是标准的MPLS + BGP结构，一样的BGP free Core效果，
只是用IPinIP模式，MPLS就可以省了。但是在MPLS方案里面PE还是有全路由，所以LISP
最重要的就是Map－Request功能。
之所以说这个更象H323查询过程，而不是DNS，原因有两个：
1. 系统里面已经有DNS，再搞一个会混淆
2. DNS没有做EID-RLOC转换，还是EID的不同表达方式，而从电话号码到Gateway的映射
更类似于EID-RLOC的关系，Admission Control就不是重点了，不过加上这功能也不错
，实际上在MAP Server上是可以实现一定级别的路由策略的。
最后的问题就是Tunnel End Router选取，取PE，就恢复到原始的MPLS＋BGP方案，MAP
Cache的规模多少近似于全路由。CE会好的多，毕竟一个site的用户不会访问整个
Internet，当然DC除外。如果允许有Default RLOC的话，整个方案看起来很不错。
running MPLS, service provider can still run... 阅读全帖 |
|
s*****g
发帖数: 1055 | 25 Let me try to answer your questions, there are two types of MPLS VPNs, MPLS
L2VPN and MPLS L3VPN, MPLS L3VPN is standardized in well known RFC2547,
payload is IP, MPLS L2VPN is a general term for (Cisco)AToM, EoMPLS and
VPLS, payload is FR/HDLC but more commonly Ethernet/Ethernet VLAN.
ISPs need to run IGP in their own internal network, because MPLS LSPs are
typically among iBGP loopbacks, iBGP has to be on top of TCP/IP, of course
you can have MPLS LSPs that are purely used to transport IP tra... 阅读全帖 |
|
c********g
发帖数: 3968 | 26 先篮板，后得分
1-20-52 G.Mikan,MPL vs. ROC. 36-61
3-04-52 G.Mikan,MPL vs. Phila. 36-41
1-16-57 M.Stokes,Roch. vs.St.L. 33-31
1-31-58 B.Russell,Bos. vs. Phila. 33-32
1-06-59 B.Pettit,St.L. vs. MPL. 35-38
10-31-59 W.Cham.Phila. vsDet. 34-36
11-04-59 W. Cham.Phila. vs Syra. 40-41
11-07-59 W.Cham.Phila. vs Bos. 30-30
11-08-59 W.Cham.Phila. vs Cin. 35-32
11-10-59 W.Cham.Phila. vs NY 43-39
11-25-59 W.cham.Phila vs Bos. 35-45
11-26-59 W.Cham.Phila. vs Bos. 33-41
12-03-59 W.Cham.Phila. vs MPL. 33-41
12-13-59 W... 阅读全帖 |
|
s*****g
发帖数: 1055 | 27 The way I understand LISP:
1. Data plane is more like moving MPLS to customer edge. Today service
provider can have BGP free core by transport IP over MPLS, while LISP
extended this "tunneling" to customer's edge router without customer running MPLS, service provider can still run existing MPLS network or they can have a pure IP core but its routers don't need full internet table. LISP
encapsulation has more overhead than MPLS. However, you can not transport IP/MPLS across multiple SPs (maybe yo... 阅读全帖 |
|
s*****g
发帖数: 1055 | 28 The way I understand LISP:
1. Data plane is more like moving MPLS to customer edge. Today service
provider can have BGP free core by transport IP over MPLS, while LISP
extended this "tunneling" to customer's edge router without customer running MPLS, service provider can still run existing MPLS network or they can have a pure IP core but its routers don't need full internet table. LISP
encapsulation has more overhead than MPLS. However, you can not transport IP/MPLS across multiple SPs (maybe yo... 阅读全帖 |
|
t*******r
发帖数: 3271 | 29 回到电脑跟前了, 那些说PTX一点儿PE feature没有的, 睁大了眼睛看着
COS related features
PTX supports following types of classifiers:
dscp
dscp-ipv6
exp
ieee-802.1
ieee-802.1ad
inet-precedence
ifl based fixed classification
PTX supports max of 32 classifier tables per Type and max of 64 classifier
tables for all types together.
The following rewrite types are supported:
dscp;
dscp protocol mpls;
- Apply to IPv4 packets entering MPLS tunnel; r... 阅读全帖 |
|
z**r
发帖数: 17771 | 30 NH应该是far end PE loopback，这个肯定是从IGP得到，同时也应该在LFIB里，你不激
活ipv4 neighbor，同时看一下show mpls for？
刚才俺也做了个实验，果然不需要在add ipv4下激活ipv4 neighbor，只需要在add
ipv6下激活ipv4 neighbor就可以送ipv6扩展了。
R1----R2----R3----R4
R1 and R4 are dual stack, R2 and R3 are IPv4 only, LDP enabled on all
interface
相关配置如下
!
!
!
ip cef
no ip domain lookup
ipv6 unicast-routing
ipv6 cef
!
!
interface Loopback0
ip address 10.1.1.1 255.255.255.255
!
interface Loopback1
no ip address
ipv6 address 2001::1/128
!
interface Ethernet0/0
no ip address
... 阅读全帖 |
|
x*********n
发帖数: 28013 | 31 比如。我现在
ping x。x。x。x r 1000次。没有drop
但是traceroute
4 216.80.5.50 4 msec 4 msec 4 msec
5 63.237.42.129 8 msec 4 msec 8 msec
6 205.171.139.62 [MPLS: Label 797744 Exp 0] 4 msec 4 msec 8 msec
7 * * *
8 67.14.8.194 4 msec 4 msec 8 msec
9 63.146.26.218 4 msec 4 msec 8 msec
10 12.122.84.54 [MPLS: Label 17124 Exp 0] 32 msec 28 msec 32 msec
11 12.122.2.206 [MPLS: Labels 17694/17132 Exp 0] 32 msec 28 msec 32 msec
12 12.122.2.126 [MPLS: Labels 23426/17132 Exp 0] 32 msec 32 msec 32 msec
中间的* * *能表... 阅读全帖 |
|
z**r
发帖数: 17771 | 32 PBT stands for Prvider Backbone Transportation, 从字面看，它不是一个给企业用
的东西，企业对于carrier用什么技术完全是不知道的，对他来说完全是透明的。
俺上一个例子中只有PE之后才是PBT，最终用户依然是dsl or fiber，对于STB来说，就
像俺上面所说，依然完全透明。
刚开始俺觉得这个PBT的确很新鲜，但是隐隐约约觉得PBT能做的，现有设备都能做，虽
然从单一的ethernet service看貌似便宜一些，但是一个carrier尤其Tier 1/2这些，
不可能只提供ethernet，PBT则根本不可能提供PPP/FR/HDLC等等给最终用户，而如果
backbone是MPLS则情况大大不同了。所以从TCO的角度看，PBT不见得就便宜，因为为了
提供PPP/FR/HDLC/ATM等业务，carrier还不得不继续维护另一套网络。如果是MPLS则没
有这个必要了。
俺觉得PBT并不是根传统的ethernet竞争，而是根MPLS，T-MPLS这些竞争 |
|
x*********n
发帖数: 28013 | 33 客户说HQ和其中4个branch连不上了，我看了一下，是升级加新的site的时候，VPN
tunnel miss掉了，或者是WAN IP改掉了，这样2个WAN IP，是ping不通的。
我弄好后email跟客户说，你的VPN修好了，结果项目经理看到了，跟我的team leader
说这个人怎么回事啊，bla。。。bla。。。
其实项目经理在sales的时候是吹的天花乱坠的，MPLS啥的都扯上了，MPLS我们压根没
有，这样我VPN一讲，客户跑来质问我们到底是VPN还是MPLS。
最后我灵机一动，就说我们是MPLS-VPN好了。。。
就这点狗屁事，又把我教育了一番，以后和客户说话，啥都别讲，只管要做好你的事情
。。。。bla。。。bla。。。
有的时候想想我没错啊。。。也有的时候想想也许每个公司都是这样。。再想想如果当
时不是这个我讨厌的老板给我offer，我就没新车，没钱住one bedroom，还和另外3人
合租在一起，住在客厅里。
哎。我真是本版抱怨之王啊。 |
|
z**r
发帖数: 17771 | 34 I agree with you on the second point. for the first point, if you call any
addtional header MPLS like, you could, but I would say the biggest advantage
of LISP is dynamic tunnel from CPE to CPE
running MPLS, service provider can still run existing MPLS network or they
can have a pure IP core but its routers don't need full internet table. LISP
IP/MPLS across multiple SPs (maybe you can with the help of BGP, but that
capability is mostly for inter-AS/CsC, it won't be scale for internet
routing, ... 阅读全帖 |
|
z**r
发帖数: 17771 | 35 I agree with you on the second point. for the first point, if you call any
addtional header MPLS like, you could, but I would say the biggest advantage
of LISP is dynamic tunnel from CPE to CPE
running MPLS, service provider can still run existing MPLS network or they
can have a pure IP core but its routers don't need full internet table. LISP
IP/MPLS across multiple SPs (maybe you can with the help of BGP, but that
capability is mostly for inter-AS/CsC, it won't be scale for internet
routing, ... 阅读全帖 |
|
|
u*v
发帖数: 18 | 37 正是这样的，这些都应该在PE上处理的。现在的问题是，traffic在Ingress的线卡上就根
据需要进行了IP分类，然后通过背板，在进入MPLS处理块对IP分类进行到EXP的映射及加
标记等处理，后面就是MPLS的机制了。当然理论上说，IP分类完全可以在CE或之前就完成
，但要记住，PE是运营商的设备，而CE完全可以不是运营商的设备，运营商不可能信任客
户的IP分类，它必须在PE入口根据SLA再做一次IP分类。另外我同意前面的一些说法，对非MPLS的情况，在Ingress这样处理有时也是很必要的，QOS很多时候在线卡上就完成了，实际上你应该把ROUTER，特别是大的象GSR，CRS-1，T640，看成线卡（里面包括线卡控制，分布式路由表等其它东西）、核心路由表维护、核心交换、核心控制等功能块独立组成，这样就不难理解了。 |
|
z**r
发帖数: 17771 | 38 ☆─────────────────────────────────────☆
zher (民工.呛呛嗟) 于 (Thu Apr 20 15:06:16 2006) 提到:
这个egress qos比较好理解，这个ingress qos，尤其ingress queueing，应该如何理解
？
☆─────────────────────────────────────☆
urv (U-R-V) 于 (Fri Apr 21 15:35:18 2006) 提到:
俺对QoS研究很少，我想至少在MPLS TE有用，对进来的Traffic进行分类，确认打什么标
签，走哪条路经，以后就是标记交换了，所以必须在进MPLS时就打上tag。
☆─────────────────────────────────────☆
zher (民工.呛呛嗟) 于 (Sat Apr 22 15:15:05 2006) 提到:
MPLW EXP bits都是在做mpls imposition的时候，要么从cos bits，要么从ip
precedence要么从dscp copy 过去的， |
|
c*****i
发帖数: 631 | 39 pe-pe的rip是igp。 ibgp是mp-ibgp,传mpls vpn route/ lable用的。当然如果你有
full mesh的ipv ibgp，就不需要rip的。
话说mpls support rip as igp吗。
上次还被人问了个TE需不需要mpls,汗阿。 |
|
z**r
发帖数: 17771 | 40 mpls不在乎是什么routing protocol啊，只是如果你需要做TE，那么得是link state
IGP。
TE的确可以不要MPLS在core里跑，只要做full meshed PE-PE TE tunnels，就可以，当
然PE上还是要enable mpls的 |
|
s*****g
发帖数: 1055 | 41 So you are saying that if all core routers are running iBGP full mesh, then
you don't need IGP? Then we are talking about the same thing, and I am
telling you that you are wrong. ;-)
Of course you need to configure MPLS in the core (define configure MPLS?), TE can establish the LSP you want, but you still need MPLS functionality to switch the labeled packets.
Guys, guys, networking 102
mesh。 |
|
z**r
发帖数: 17771 | 42
then
TE can establish the LSP you want, but you still need MPLS functionality to
switch the labeled packets.
configuring mpls means enable ldp right? if you have full meshed bi-
directonal TE tunnels for each VPN, the P routers don't have to have LDP
tuned on.
I just checked this in "mpls fundamentals", page 321, and I am posting the
content here:
TE Tunnels Between PE Routers
When two TE tunnels (one for each direction) exist between a pair of PE
routers and the Border Gateway Protocol (BGP) n... 阅读全帖 |
|
s*****g
发帖数: 1055 | 43 BGP send-label is specifically designed for inter-AS MPLS VPN case, not used
for MPLS label distribution for general MPLS LSP establishment purpose, LDP
can advertise BGP routes if it is configured to do so, but ... but .... are
you guys talking about basic networking principle or TAI(2)GANG(4) ing? |
|
s*****g
发帖数: 1055 | 44 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 | 45 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 |
|
t*******r
发帖数: 3271 | 46 来自主题: EmergingNetworking版 - 一个面试题 以下回答不针对楼主提的问题, 只是一般性的看法:
1, 最好不要用私有地址做PE-P-PE的互联, 因为一般是一张网干多件事, 而且loopback
是私有IP地址会有
管理的问题;
2, 同一个VRF里做Internet访问确实比较扯, 早年JUNIPER实现过, 十分不scalable.
用户是公网地址
那是好的, 用户是私有IP让你做VRF里的NAT那才是生不如死;
3, Saiwong提出的2个IFL(逻辑接口)确实是个好办法, 但是, 仍然需要SP端做相对复杂
的配置更改;
4, 对于P router(P only), 的确要减轻控制平面的压力, 最好是RR/P/PE功能性分开.
细化一点儿来说,
L3VPN RR/L2VPN(VPLS) RR/IPv4 RR/IPv6 RR最好也分开, 当然, PE就不必了.
BGP table还是要适当控制(route-target方式), 避免所有PE(不需要full bgp feed的)
都有全网路由
针对快速无缝切换:
A. 先试点开几个VPN site的业务(不同类的, 比如VPLS/L3VPN), 证明端到端LSP是通的
, V... 阅读全帖 |
|
c******0
发帖数: 881 | 47 从上个月1号开始正式找工作到现在已经一个月有余，晚睡早起，接发邮件，接打电话
，看书复习，GNS练手，说辛苦还真是挺辛苦的，不过好在总算慢慢的开始有结果了。
昨天今天各拿一个offer，虽然还有公司在继续走流程，但是这两个offer给的考虑时间
比较有限，加上也累了，想考虑着接受一个开始认真工作了。
大大小小的公司各种面试已经数不过来了，最多的一天连面四个技术面，加上HR的电话
，总共打了5个多小时。这其中能叫的上名字的，也就是那么几个公司cisco，juniper
，amazon，AT&T，还有一些pandora之类看起来貌似有一些规模的公司。
CISCO，面试题目最理论的公司。其实这是我没有想到的。跟juniper的面试风格截然相
反，
按理说应该得有一些经验相关的问题吧，而且要求上也是有要求工作经验的。总共三个
工程师
面，问的问题全部很理论很标准，类似ospf邻居建立过程，ospf E1/E1/N1/N2的区别，
bgp邻居建立过程，bgp包的类型总类等等，基本上对答如流，毫无压力。感觉应该是能
进入下一
轮的，但是时间战线还是一如继让的拉得很长，估计是没有时间等了。
AMAZ... 阅读全帖 |
|
|
|
|
