w***a
发帖数: 29 | 1 问一下有没有办法改 现有软件（本身不支持DSCP)发出的包的DSCP? 我知道group
policy 里面可以定义
DSCP tag value. 但是不知道可不可以把它和应用程序绑定起来。 |
|
t*******r
发帖数: 3271 | 2 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 |
|
x*********n
发帖数: 28013 | 3 coworker有个case，就是Cannot Apply "service-policy output " on
the Interface
然后呢，TAC回复
I believe the egress service policy is limited to matching based on DSCP
values, while you have more options for ingress. Queueing evaluates the
internal DSCP and determines which of the four egress queues in which to
place the packet. The DSCP value is mapped to a CoS value, which selects one
of the queues.
Please configure a new policy-map used for egress interface matching only
DSCP values and remove... 阅读全帖 |
|
b******s
发帖数: 5329 | 4 af31 = dscp 26,
cs3 = dscp 24,
Cisco过去用af31做signal QoS，现在改用cs3了。当然这些都是名字，具体要看DSCP
的数值，不过既然已经有约定俗称，大家基本还是遵守的。 |
|
z**r
发帖数: 17771 | 5
这个概念不对啊。classfication哪里都得要。policing也可以，只不过因为policing
的特殊性，既然要drop某些packets，那么就最好干脆在ingress drop，省得router处
理了半天再drop。marking也是都可以。
QoS和硬件息息相关，所以撇开硬件平台，只能谈论generic的。
qwest是sp，通常不信任customer的dscp/cos/ipp，所所以他们肯定会mark自己的东西
，但同时都会preserve customer dscp/cos/ipp等，如果没有preserve，估计这样的
service比较便宜。同样作为客户，你们的网络也有自己的qos要求，如果你们自己的
dscp/cos/ipp等没有被qwest preserve，那你们自己就得在ingress做marking来符合你
们自己网络的要求 |
|
z**r
发帖数: 17771 | 6 AF31=DSCP 26
CS3=DSCP 24
怎么就一样了？
RFC 4594 is what you need to read |
|
t*******r
发帖数: 3271 | 7 楼歪了哈. 反正, 你最后明白那些alias代表什么就行, 具体的就不争论了.
贴一个猪家的基本符合RFC规范的alias定义
tonyblair@kurt-re1> show class-of-service code-point-aliases inet-precedence
Code point type: inet-precedence
Alias Bit pattern
af11 001
af21 010
af31 011
af41 100
be 000
cs6 110
cs7 111
ef 101
nc1... 阅读全帖 |
|
h*****h
发帖数: 1392 | 8 敝公司位於矽谷Sunnyvale，是美國公司，我們現請人做QA Engineer。我們是做home
router/modem的，公司狀況非常好，工作穩定。應聘者必須有足夠網絡方面的知識，並
且勤勉敬業。年資要求不高。有1-3年經驗就好。資深者請另覓他家。New grads如果確
實優秀，也會考慮。公司有Medical/Vision/Dental Insurance。辦H1和綠卡。資深人
士不適合。不能接受平時和週末偶爾加班的人士也不要申請了。根據資歷和knowledge
，時薪從20到25美金不等。有意者請用站內郵箱、站內短信聯絡，或發email到hshensc
@gmail.com。拜託謝謝！
Network QA Engineer
We are looking for a self-motivated and detail-oriented Network QA Engineer.
The candidate will be responsible for testing our networking devices and
will be involved in all aspe... 阅读全帖 |
|
|
z**r
发帖数: 17771 | 10 what's the problem? the broadcast dscp request? a helper address should work
for it, right?
to |
|
z**r
发帖数: 17771 | 11 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，当然也可以人工set，但是都是发生在ingress
traffic已经进到了router以后。当然俺说的这个市ip2mpls，在做swap的时候，也可以
classify based on exp bits，但是对于ingress qos来说，只是classification的不同
，并没有对queueing产生
任何影响
标 |
|
z**r
发帖数: 17771 | 12 ☆─────────────────────────────────────☆
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， |
|
s*****g
发帖数: 1055 | 13 来自主题: EmergingNetworking版 - 网络技术问题 Not at all, when the packet is routed, L2 header are stripped, lot of IP
header fields can be changed, router will decrement IP TTL, depends on
configuration, TOS/DSCP bit can be modified, if it is a NAT device, source/
destination IP addresses can be changed, etc, etc, outbound L2 encap can be
PPP, without IP checksum, how does next hop router know the IP packets'
integrity are maintained?
The same argument applies to why we need TCP/UDP checksum, the whole point of layered TCP/IP is that upp |
|
x*********n
发帖数: 28013 | 14 书上说EF，然后AF，然后CS，但是dscp的value是一样的。
还有人说AF是cisco弄出来的，一般都是CS。
我觉得很蹊跷啊。 |
|
t*******r
发帖数: 3271 | 15 和别的协议规定不同, 针对dscp/exp/ip precedence这玩意儿, RFC也只是建议而已,
alias的名字你完全可以不用rfc4594上规定的, 只要你自己的网络内部CoS策略一致即
可.
举例:
CS4=100000, 代表实时的高优先级流量
CS1=001000, 代表低优先级流量
你说我部署和做配置的时候, 把code points互换行不行?
行呀!!! 只要你网络内部code points的设置使得同一类别的流量的CoS策略一致, 当然
可以, 根本没人管你alias叫什么. 这个RFC里面的图3只是个建议. |
|
m**t
发帖数: 1292 | 16 理论上讲，有人做COS remarking。进来一个DSCP value， 出去用另外的value |
|
x*********n
发帖数: 28013 | 17 RSVP我以为没人用了呢？现在不是都是DSCP么。 |
|
a***n
发帖数: 262 | 18 backbone are all 10G, to building is 1G, access is 10/100/1000.
I will have Avaya Team come to evaluation UC/VoIP readiness
next week. We will start with trust dscp end-to-end first, then
go to priority queue for UC/VoIP if needed. Other class of traffic
will have to wait for later consideration.
backbone links are WS-6704 WS6708 WS6716 10G. egress mostly 1p7q4t.
1Gs are almost 1p3q4t. |
|
x*********n
发帖数: 28013 | 19 queuing is more like egress thing in my mind, and police-map is for remark
or dscp map function,
let me know if I am wrong, not sure about those stuff... |
|
z**r
发帖数: 17771 | 20 IntServe
- RSVP
DiffServe
- Classification, IPP, COS, DSCP, IPv6 TC, MPLS
- Marking
-Explicit marking
-Implicit marking, policing set, trust domain, mpls modes, etc.
- Congestion management, PQ, CQ, LLQ, CBWFQ, scheduler, etc.
- Congestion avoidence, RED, WRED, FRED, TCP window
- Shaping and policing, token bucket, strict policer, conditional policer,
etc.
- MQC, H-QoS
- Implementation guide |
|
x*********n
发帖数: 28013 | 21 queuing is congestion management tool
shaping and policing is congestion avoidance tool
shaping is not just drop, it is actually buffer the packets.however,
policing is.
WRED has function such as mimimum threshold, and maximum threshold, between
them, there is probability to mark all the packets, based on dscp or ip
precedence, the chance of each packet to be drop is different,sounds like
smart, right?
I wish i could had chance to listen to zher's lecture, some topics I even
haven't heard of bef... 阅读全帖 |
|
z**r
发帖数: 17771 | 22 2 things you are missing:
1. there are always multiple tcp sessions in the real world
2. without congestion avoidance like RED, whenever there is a drop, it's
tail drop.
So, when the link is congested, all the packets that are not in the queue
yet will be tail dropped, and these packets are from multiple tcp sessions.
Thus results in transmission wave or global sync, big waste of bandwidth (
queue actually)
With congestion avoidance like RED/WRED/FRED/etc., the lower priority
packets with higher... 阅读全帖 |
|
