Sunteți pe pagina 1din 4

ABACUS Call Centre Network Adudit Analysis Problem Point of Failure 1)If OSN and SW.Abacus-01.

L3 @ OSP2 go down at same time, there is a backup Radio Link available. However, this backup cant be used for redundancy due to network design in place. Impact is High Missing/Inappropriate QOS 1)No or Inappropriate DSCP (Differentiated Services Code Points) and COS (Class of Service) Marking is configured at point of traffic origination/termination. Impact is High.

Interface Errors 1)Huge amount of output packet drops are observed specifically on Gig 0/1 of SW.Abacus-01.L3 and LLR020-SW01 Gigabit 1/0/7 which is uplink between OSP2 and TP, and at few other interfaces of SW.Abacus02.L3 and LLR020-SW02 which are serving as backup links between OSP2 and TP. Highest peak bandwidth utilization monitored during Audit duration is 53 Mbps DL and 14 Mbps UL on NMS (Solarwinds NPM) at OSP2 site. Impact is High. Layer 3 Interface Higher Broadcast & Multicast Packets There is large number of Broadcast & Multicast packets received on many L3 interfaces of switches (e.g.: SW.Abacus-01.L3 Gigabit 0/1 0/2 0/9 0/10, LLR020-SW01 Gigabit 1/0/7, etc.). Impact is medium.

Layer 3 Interface Flapping: Link flaps are observed on: - LLR020-SW01 on May 27, 2011 @ 4:21:08 resulting in broken OSPF neighborship at the same time. The link recovered at May 27, 2011 @ 4:25:50 resuming OSPF neighbourship at 4:28:14 - LLR020-SW02 on May 27, 2011 @ 4:27:02 resulting in broken OSPF neighborship at the same time. The link recovered at May 27, 2011 @ 4:27:13 resuming OSPF neighbourship at 4:28:18 Duration for which OSPF neighborship was broken at both switches on Fiber link was for 1 minute and 11 seconds. The radio link may have taken up the traffic load but Fiber being prime media is suspected down. In another case, Radio Link interface Gigabit 1/0/11 at LLR020-SW01 flapped many times on June 6th, 7th, 17th, and 18th 2011 that lead to broken OSPF neighborship. A similar trend was observed and noted at TP Switches in OSP2. Note: NTP clock is out of sync that was causing difficult at the point of analysis. Please refer to [5.1.6] and many ports are configured on Auto Speed and Duplex. Impact is high. NTP Clock Drifted NTP clock of LLR020-SW01 & LLR020-SW02 with SW.ABACUS-01.L3 & SW.ABACUS-02.L3 drifts by +1 Hour approximately. PE-CE OSPF Area 4001 is configured on LLR020-SW01, LLR020-SW02, SW.Abacus01.L3, and SW.Abacus-02.L3 as normal area. IOS Upgrade: TP Switches are using a bit old IOS version while newer version is available since May 2011

e Network Adudit Analysis Remedy 1) Redudant OSN can be placed and OF link can be terminated on separate OSN. 2) OF Link connected to S-Core switch with 70 MB BW can be upgraded to 100 MB. 3) For Secondary link we can provide MW link Protection as well.

Qos Marking Required. Both QOS and COS Marking should be configured on uplinks (L2, L3). 1)Ethernet Cable connectivity should be verified for physical damages. 2. DUPLEX/SPEED should be set to fixed values. Recommended fixed values are actually the best (e.g. 1000 Mbps speed, full duplex) combination that can be mutually agreed between two ports. We need to Confirm that Switch ports are autonegotiated. 3)Switch port modes should be statically set. All uplinks should be configured as static 802.1Q trunks (NNI) and all PC/System connections (UNI) should be configured as access ports in a particular VLAN. 4. Verify maximum bandwidth pipe available between OSP2 and TP in SDH Core. If this value turns out not to be greater than at least 53 Mbps, same should be increased.

1) Storm Control feature for broadcast and multicast traffic should be configured.

1)None, if planned activity. 2. If not, OSN connectivity and Fiber Transmission path needs to checked. 3. DUPLEX/SPEED should be set to fixed values. Recommended fixed values are actually the best (e.g. 1000 Mbps speed, full duplex) combination that can be mutually agreed between two ports. Confirm that switch ports can autonegotiate. 1)Clock should be re-synchronized with appropriate NTP Server on relevant switches. Area 4001 should be configured as Not So Totally Stub. ACL can be configured on Core switches so that CC user can only access the respective destiantions.

Upgrade IOS

S-ar putea să vă placă și