cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
6393
Views
0
Helpful
9
Replies

Upgrade core switch without disrupting network

VAbr AVib
Level 1
Level 1

I am planning to upgrade the current core switch(3750) to 6509 series switch. Since we have a production network running we have to plan for an online core switch upgrade.

Could you please help me with the high level steps to acheive this without any kind of disruptions.

Thanks.

9 Replies 9

patrick.preuss
Level 1
Level 1

Hi

Depends on the links you can place to The Access. Systems connected direct will have an interrupt.

1) Review The l3 . Is a fhrp configured? How is The Wan connectivit?
If possible implement an fhrp.
2) Review spanning tree protocol? If possible use Rapid pvst.
3) implement an uplink between Old and New Core.
4) if possible Setup The Access as a triangle.
5) Place New cables from The New Core to The systems that Need Hard to be rewired so The Time to Place The Patch / Connection is minimised.
6) Check and migrate l3
7) Check and migrate L2

Sent from Cisco Technical Support iPhone App

1. I am not familiar with FHRP, its the First Hop Redundancy Protocol but i am not sure how it differ from HSRP and also how it helps in a switch level.

2. Apart from that what you suggest is to check the Spanning tree, again i  need to understand more about Rapid PVST

3. Yes, that is anyway there in the plan. Does the new switch be made as VTP server before trunking ?

4. All access switched to be connected in triangle you mean star topology ?

5. Connect systems to new core, you mean connect access switches(step.4) to new core and the put systems there?

6. Migrate L3: what additional tasks you think we have with this this step?

7. Migrate L2: assuming that this is only the uplinks and trunks to VTP clients ? Anymore configurations ?

My plan is not to add any new feature as of now. Rather plan for a smooth migration and then once we migrate and release the current core we can then thinkg about many more options that 6500 series switch can provide. Anyway i am giving below the core switch configurations for your reference.

CORE-SWITCH# show version

Cisco IOS Software, C3750 Software (C3750-IPBASE-M), Version 12.2(25)SEB4, RELEASE SOFTWARE (fc1)

Copyright (c) 1986-2005 by Cisco Systems, Inc.

Compiled Tue 30-Aug-05 15:47 by yenanh

ROM: Bootstrap program is C3750 boot loader

BOOTLDR: C3750 Boot Loader (C3750-HBOOT-M) Version 12.2(25r)SEC, RELEASE SOFTWAR

E (fc4)

CORE-SWITCH uptime is 2 weeks, 1 day, 23 hours, 35 minutes

System returned to ROM by power-on

System image file is "flash:c3750-ipbase-mz.122-25.SEB4/c3750-ipbase-mz.122-25.S

EB4.bin"

cisco WS-C3750-48TS (PowerPC405) processor (revision M0) with 118784K/12280K byt

es of memory.

Processor board ID CAT0950Z4YN

Last reset from power-on

43 Virtual Ethernet interfaces

96 FastEthernet interfaces

8 Gigabit Ethernet interfaces

The password-recovery mechanism is enabled.

512K bytes of flash-simulated non-volatile configuration memory.

Base ethernet MAC Address : 00:16:9D:32:18:00

Motherboard assembly number : 73-9680-09

Power supply part number : 341-0028-02

Motherboard serial number : CAT095106P5

Power supply serial number : DTH09481VMJ

Model revision number : M0

Motherboard revision number : A0

Model number : WS-C3750-48TS-S

System serial number : CAT0950Z4YN

SFP Module assembly part number : 73-7757-03

SFP Module revision Number : A0

SFP Module serial number : CAT09500L58

Top Assembly Part Number : 800-25854-02

Top Assembly Revision Number : C0

Version ID : V05

CLEI Code Number : CNMV300CRE

Hardware Board Revision Number : 0x01

Switch Ports Model SW Version SW Image

------ ----- ----- ---------- ----------

* 1 52 WS-C3750-48TS 12.2(25)SEB4 C3750-IPBASE-M

2 52 WS-C3750-48P 12.2(25)SEB4 C3750-IPBASE-M

Switch 02

---------

Switch Uptime : 2 weeks, 1 day, 23 hours, 30 minutes

Base ethernet MAC Address : 00:16:9D:42:70:80

Motherboard assembly number : 73-9675-09

Power supply part number : 341-0029-04

Motherboard serial number : CAT0951078C

Power supply serial number : LIT094806FZ

Model revision number : J0

Motherboard revision number : A0

Model number : WS-C3750-48PS-S

System serial number : CAT0950R5FL

SFP Module assembly part number : 73-7757-03

SFP Module revision number : A0

SFP Module serial number : CAT095108TG

Top assembly part number : 800-25858-03

Top assembly revision number : D0

Version ID : V05

CLEI Code Number : COM1W00ARB

Configuration register is 0xF

CORE-SWITCH#sh vtp status

VTP Version                     : 2

Configuration Revision          : 138

Maximum VLANs supported locally : 1005

Number of existing VLANs        : 30

VTP Operating Mode              : Server

VTP Domain Name                 : CITYLAN

VTP Pruning Mode                : Enabled

VTP V2 Mode                     : Disabled

VTP Traps Generation            : Disabled

MD5 digest                      : 0x0E 0x40 0x8C 0x2E 0xAC 0x79 0xB7 0x2A

Configuration last modified by 192.168.1.1 at 12-12-12 06:52:10

Local updater ID is 192.168.1.1 on interface Vl1 (lowest numbered VLAN interface

found)

1. I am not familiar with FHRP, its the First Hop Redundancy Protocol but i am not sure how it differ from HSRP and also how it helps in a switch level.

FHRP are the class of First Hop Protocols like HSRP, VRRP, GLBP and so on.

If your 3750 are doing Routing Functions (They can you have installed IP Base)

You should plan first the change of the first hop and the routing.

2. Apart from that what you suggest is to check the Spanning tree, again i  need to understand more about Rapid PVST

RPVST is Rapid Per VLAN Spanning Tree

The Transition from Blocking is a lot faster than in classic spanning tree.

3. Yes, that is anyway there in the plan. Does the new switch be made as VTP server before trunking ?

as far as i know not. Please test it here you can have more than one VTP server.

4. All access switched to be connected in triangle you mean star topology ?

If you take you core and the new core and place an access switch there with a link to old and new it looks like a triangle.

if you can cable this way, it would be the easiest way, depends on you topologie.

This setup will minimise outtages during remove of the cables. from the old one.

5. Connect systems to new core, you mean connect access switches(step.4) to new core and the put systems there?

Not Exachly prepare the Change, trow new cables and so on (Cabling is the worst part)

6. Migrate L3: what additional tasks you think we have with this this step?

When you are using routing on the catalyst change over those functions here.

7. Migrate L2: assuming that this is only the uplinks and trunks to VTP clients ? Anymore configurations ?

Configuration is done allready, this means plug the cables in the right places:-)

Leo Laohoo
Hall of Fame
Hall of Fame

Not possible because you have to physically move your copper/fibre links from the 3750 to the 6500.

sorry i didnt get your point ?

Are you saying it is not posible for an online migration ?

My only problem is with the VLAN database so i thought if i can bring another VTP server (3750) and connect to the new core)6509 then propogate the VLAN database. Later we can bring the 3750 switch offline.

If the above option is ruled out then what otehr option is there to avoid complete VLAN rework to bring 6500 switch as core ?

You can use the 6500 as VTP Server directly.

Finally I decided to simultaneously build a core switch  with same configurations of existing core switch, then physically move each access switches here. This was I can limit my planning at the cabling level itself and also I can roll back if there are issues after migration.

Hope this is the most feasible approach if we have plans to replace the core switch from 3750 to 6509.

Please let me know your thoughts.

How did this work out for you? I am doing the same thing this weekend.  any issues?

thanks

I would not recommend you doing this. If you try to build a new core switch with the same configuration with existing one and switch access switches' uplink cables one by one from old core to new core, then you will get huge network disruption. The reason for this is that as soon as you switch ONE switch' uplink cable (to the old core, if it has only one cable) to the new core, users connect to that access switch will lost connection the all your servers, Internet and WAN. And you may get the blame for doing this.

In fact Patrick's suggestions may be the best for you (the disruption to the whole network would be less then 1 second each step if planned it (and tune the timers) carefully and since you can't get the down time, you do implement one step a day to minimize the impact). If your network can't tolerate the disruption of swaping all cables from old core to new core, I would suggest postpone the upgrade and ask Patrick (or others) to explain it then make a plan for the upgrade.

Also, it's very important to let others get a better understanding of your network if you want his help. The more he knows about your network, the more risks could be avoided.

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Review Cisco Networking products for a $25 gift card