04-14-2004 03:46 AM
My client was using IBM router at remote and connecting to IBM router at HQ with WAN line speed of 64kbps. Running OSPF. The application is running on SNA protocol.Remote branch will access the "HOST - Mainframe" at HQ.
After the migration to Cisco routers the customer complain the accessing to the "HOST' became slower.
Iniatially when on IBM platform the network design was very straight forward. The remote branch directly connected to Frame-relay cloud to HQ with one PVC.
Now there is 2 units of 7206 Cisco routers at HQ for redundancy purpose.
Branch remain with frame-relay 64 kbps but using 2 logical sub interface with different PVC to connect to the 2 units of 7206 routers at HQ.
Since running on OSPF routing protocol, IP packets can loadbalance via the both PVC but how about the SNA protocol?
Will there be any issue on the amount of bandwidth being use for SNA application after the migration since the remote branch will only connect to one of the router at one point of time using DLSW to access the HOST.
Below is my remote config. Is there any issue with this config?
dlsw local-peer peer-id x.x.x.1
dlsw remote-peer 0 tcp 141.x.x.1 <hq router 7206_1 token ring interface>
dlsw remote-peer 0 tcp 141.1.243.3 backup-peer 141.1.243.1 linger 3
dlsw icanreach sap 0 4 8 C
dlsw bridge-group 1
!
interface FastEthernet0/0
ip address 141.x.x.x.255.255.0
no ip redirects
ip route-cache flow
ip ospf authentication-key xxxxx
speed 100
full-duplex
bridge-group 1
!
interface Serial0/0
no ip address
encapsulation frame-relay IETF
no ip mroute-cache
no fair-queue
cdp enable
frame-relay lmi-type ansi
!
!
interface Serial0/0.899 point-to-point
description PVC to HQ 7206_2
bandwidth 64
ip address 10.x.x.x.x.255.252
ip ospf authentication-key xxxx
ip ospf network point-to-point
frame-relay interface-dlci 899
!
interface Serial0/0.900 point-to-point
description PVC to HQ 7206_1
bandwidth 64
ip address 10.x.x.x.x.255.252
ip ospf authentication-key xxx
ip ospf network point-to-point
frame-relay interface-dlci 900
04-14-2004 05:56 AM
Based upon the information given, I would suspect what is occuring is that you do not have IP MTU path discovery turned on and since there is duplicate paths, there are out of order TCP packets between the DLSW peers. So, add this command to the DLSW routers: ip tcp path-mtu-discovery. Then, when you get an outage period, simply do a dlsw disable and then no dlsw disable on one of the dlsw routers to reset the session. I wrote a webpage to describe what this does at http://www.cisco.com/en/US/partner/tech/tk331/tk336/technologies_tech_note09186a00801d3a9d.shtml.
If you can, post a show tcp and show interface s0/0 from the routers and we can make sure. On top of doing this, we might need to implement frame-relay traffic shaping with TCP port 2065 high priority to ensure DLSW can get through. This would depend on the CIR and burst of the PVC's and how much is currently being utilized.
04-14-2004 09:50 PM
Hi,
Thanks for the information. Here show tcp command
sh tcp
Stand-alone TCP connection from host 141.1.243.1
Connection state is ESTAB, I/O status: 1, unread input bytes: 0
Local host: 141.1.209.1, Local port: 2065
Foreign host: 141.1.243.1, Foreign port: 22452
Enqueued packets for retransmit: 0, input: 0 mis-ordered: -1 (0 bytes)
TCP driver queue size 0, flow controlled FALSE
Event Timers (current time is 0x32FEC44):
Timer Starts Wakeups Next
Retrans 9338 380 0x0
TimeWait 0 0 0x0
AckHold 10783 0 0x0
SendWnd 0 0 0x0
KeepAlive 1407 0 0x0
GiveUp 8957 0 0x0
PmtuAger 0 0 0x0
DeadWait 0 0 0x0
iss: 3135737176 snduna: 3139076055 sndnxt: 3139076055 sndwnd: 20342
irs: 3287893903 rcvnxt: 3290222475 rcvwnd: 20022 delrcvwnd: 458
SRTT: 960 ms, RTTO: 4265 ms, RTV: 3305 ms, KRTT: 0 ms
minRTT: 20 ms, maxRTT: 3244 ms, ACK hold: 200 ms
Flags: passive open, higher precedence, retransmission timeout
Datagrams (max data segment is 536 bytes):
Rcvd: 22679 (out of order: 1258), with data: 11790, total data bytes: 2328571
Sent: 22348 (retransmit: 380, fastretransmit: 17), with data: 9722, total data b
ytes: 3347990
tty66, virtual tty from host 10.6.1.9
Connection state is ESTAB, I/O status: 1, unread input bytes: 0
Local host: 141.1.209.1, Local port: 23
Foreign host: 10.6.1.9, Foreign port: 54786
Enqueued packets for retransmit: 1, input: 0 mis-ordered: 0 (0 bytes)
Event Timers (current time is 0x32FF01C):
Timer Starts Wakeups Next
Retrans 43 3 0x32FF21E
TimeWait 0 0 0x0
AckHold 34 5 0x0
SendWnd 0 0 0x0
KeepAlive 15 0 0x330D368
GiveUp 0 0 0x0
PmtuAger 0 0 0x0
DeadWait 0 0 0x0
iss: 556178450 snduna: 556185105 sndnxt: 556185105 sndwnd: 4114
irs: 28892121 rcvnxt: 28892229 rcvwnd: 4021 delrcvwnd: 107
SRTT: 330 ms, RTTO: 503 ms, RTV: 173 ms, KRTT: 1006 ms
minRTT: 88 ms, maxRTT: 860 ms, ACK hold: 200 ms
Flags: passive open, higher precedence, retransmission timeout
Datagrams (max data segment is 536 bytes):
Rcvd: 82 (out of order: 15), with data: 47, total data bytes: 107
Sent: 82 (retransmit: 4, fastretransmit: 0), with data: 60, total data bytes: 7190
04-14-2004 09:53 PM
Hi,
Here is the show interface command
2620#sh int s0/0
Serial0/0 is up, line protocol is up
Hardware is PowerQUICC Serial
MTU 1500 bytes, BW 1544 Kbit, DLY 20000 usec,
reliability 255/255, txload 1/255, rxload 1/255
Encapsulation FRAME-RELAY IETF, loopback not set
Keepalive set (10 sec)
LMI enq sent 1237, LMI stat recvd 1236, LMI upd recvd 0, DTE LMI up
LMI enq recvd 0, LMI stat sent 0, LMI upd sent 0
LMI DLCI 0 LMI type is ANSI Annex D frame relay DTE
Broadcast queue 0/64, broadcasts sent/dropped 3737/2, interface broadcasts 330
5
Last input 00:00:00, output 00:00:00, output hang never
Last clearing of "show interface" counters 14:54:37
Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 0
Queueing strategy: fifo
Output queue: 0/40 (size/max)
5 minute input rate 1000 bits/sec, 2 packets/sec
5 minute output rate 0 bits/sec, 0 packets/sec
26362 packets input, 3450007 bytes, 0 no buffer
Received 0 broadcasts, 0 runts, 0 giants, 0 throttles
0 input errors, 0 CRC, 0 frame, 0 overrun, 0 ignored, 0 abort
17843 packets output, 3140527 bytes, 0 underruns
0 output errors, 0 collisions, 3 interface resets
0 output buffer failures, 0 output buffers swapped out
15 carrier transitions
DCD=up DSR=up DTR=up RTS=up CTS=up
I've implemented traffic shaping at the routers as below but the frame-relay packets keep dropping.
interface Serial0/0
frame-relay class JPN_QOS_64K
frame-relay traffic-shaping
!
map-class frame-relay JPN_QOS_64K
frame-relay traffic-rate 32000 64000
no frame-relay adaptive-shaping
frame-relay custom-queue-list 1
logging trap warnings
logging 155.2.1.156
access-list 100 permit tcp any any eq 2065
access-list 101 permit ip 141.1.209.0 0.0.0.255 141.1.243.0 0.0.0.255
queue-list 1 protocol ip 1 list 100
queue-list 1 default 5
queue-list 1 queue 1 byte-count 3000
dialer-list 1 protocol ip permit
04-15-2004 04:40 AM
Ok, from your show tcp, the 141.1.209.1 to 141.1.243.1 show as extremely latent. In the show tcp, the Smooth round trip time is 960 ms or .96 sec and the maximum round trip time was 3244 ms or 3.244 secs. So there is an IP delay problem on this connection. I do advise to implement the path mtu discovery and reset the tcp links to pick up the new maximum segment size and this will help alot.
For the traffic shaping, do it like this:
interface s0
encapsulation frame-relay
interface s0.1 point
frame-relay class cisco
Then,
map-class frame-relay cisco
frame-relay mincir 32000
frame-relay cir 64000
frame-relay priority-group 1
Then,
priority-list 1 protocol ip high tcp 2065
All of this will help out on the router side but I think there could be something else occuring on the wan to make the delay that significant.
04-16-2004 12:58 AM
Hi,
Really appreciate your help.
Just to confirm,
Do i need to put the path mtu discovery only on the LAN interfaces which involve in DLSW or WAN interfaces as well?
By applying the traffic shaping on my 2 sub interfaces at my remote routers, will the bandwidth not being fully utilise since remote router will be only connecting to one of the HQ router at one point of time. Since the mincir is 32000 both subinterfaces, will it only use 32kbps bandwidth for the connection or it can burst to 64kbps?
04-16-2004 03:23 AM
The IP mtu path discovery is a global command.
You will shape to a 32K CIR and 64K burst with the config I did.
04-17-2004 08:50 PM
Hi,
Don't we need to enable "frame-relay traffic-shaping" on the the interface s0?
It's is missing from your sample config.
I need to enable traffic-shaping on both (HQ & remote) routers, right?
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide