Thursday, February 28, 2013
WAN technologies
WAN
Requirements are
SLA
bandwidth
latency
loss
Cost
utilization.
Goals and policies
Applications and growth
budget
ISDN
BRI - 128kbps
PRI is 1.54 Mbps like a T1.
DSL mainly Async DSL
Cable Cable Modem Termination System CMTS
Wireless
Bridge is point ot point
LAN is 150 - 300 Mbps max
MObile GSM , GPRS ,, UMTS 3G , LTE
FrameRelay.
PVC permanent
SVC switched just for the duration.
Sonet Circuit baed.
OC 1 51
OC 3 155
OC 12
OC 48
MPLS
Goal is to use labels instead of routing tables.
CE--> PE--> P __> p --> PE--> CE
Dark Fiber is Point to Point
DWDM runs on the SONET and increases the ability to split ligh Lambdas.
CIR Commited infra
Burst above
Same PPDIOO
Prepare
plan Desgin
implement operate
optimize.
Voice 250 ms
Video can afford to lose some pakcets voice CAN'T
Private Dark Fiber
Leased TDM sonet
Shared MPLS framerelay.
Fixed cost equipment and setup
recurring are the monthly charges.
QoS
classify it based on protocol or a matching ACL.
Classification orr coloring.
Congestion management
Link Efficiecy for low speed to reduce latency and jitter
traffic shaping use ingress and egress flows.
Classification.
Identify and mark. = classification
NBAR can use layer 7 to recognize applications and then label them with a priority.
CAR commited access rate. -- Basically match a traffic and then apply a priority
Congestion management
FIFO first in first out
PQ priority queiuing sets up 4 levels.
however LEVEL High must be emptied before any of the other three are allowed.
So if you have too much traffic at level High , the other traffic will suffer badly.
Custom Queuing 16 Queues, you can set up the queue limit.
so when it is full. The other queus get serviced.
WFQ - this is for interfaces below 2.0 Mbps .
it creates two queues - high bandwidth. low bandwidth.
low gets service first.
Class Based Weighted Fair Queiuing.
SAme as WFQ but you can use access list and it is modular.
LLQ
Low Latency Queuing.
Basically class based weighted fair queuing + PQ.
So voice EF gets serviced first + there is a limit on it.
Traffic Shaping.
Slowing down traffic so the utilization is high but consistent.
helps to curb bursts into longer flows.
Policing.
Dropping traffic that passes a limit.
'Link Efficiency.
LFI link fragmentation and interleaving - breaks up large packet flows and puts small voip in between
MLP multilink PPP bonds links like 2*T1
RTP real time transport. Compress the RTP header from 40 bytes to 2-5 Bytes.
Window Size.
Extend the Window size to allow more data in before you have to send an acknowledge.
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment