InterVlan routing problem with Multicast IP( Exchange CAS array IP)

[sean chang]

Thank you everyone for helping me out, I come from Oracle DBA backround so I need and do get lot of helps  from experts here.

My switch is: catalyst 3560X with 122(53)SE IPBASE.

here is some of my Vlan configurations:

I don't use  ROUTE Port, I configure switch like this:

Vlan 168:  192.168.1.1 /24

Vlan 1: 10.1.1.1 /16

IP route 0.0.0.0  0.0.0.0 10.1.1.254   ( which is PIX 515E)

All PCs in LAN 168 can go Vlan 1 servers and printers, except our Exchange 2010 CAS array. The CAS array uses Windows 2008 Network load balance (NLB) multicast, PCs in Lan 168 can use individual CAS no prblem, but  they can't use the cluster IP of the CAS array, I read documentation Multicast route needs IP SERVICES feature, which we don't have now, but I only need a routing path from another Vlan to this Cluster IP ( which consists 2 CAS servers), is there any way to accomplish such tasks?  Any help and advises are greatly appreciated.

[Souvik Ghosh · Accepted solution]

Hi,

You would need to confiugre a static ARP on the L3 device ( device doing the routing ) mapping the multicast IP to the multicast MAC and on your switch you would need to configure a static mac for the multicast mac for all the ports where the cluster devices are connected.

follow the link below for detail read.

http://www.cisco.com/en/US/products/hw/switches/ps708/products_configuration_example09186a0080a07203.shtml

regards,

Souvik

[sean chang]

Thank you Souvik,

Your advice is almost the solution to my problem, but my enviroment here is a little complicated:  we only use Virutal Machines here, one CAS is on Windows 2008 R2 Hyper-V, one CAS is on Vmware  ESXi 4.1 , both Virtual machines are running Windows 2008 R2 ,  I could trace their physcial connection ports to this Switch, but they are shared  by dozens of other Virtual Machines , my question  is  :

#1---Does this  static ARP configuration will affect other VMs which use the same switch interface?

#2-- I checked one of this CAS has MAC addess 0015.5D02.0804 ( from  Hyper-V manager setting), should I use this MAC for static ARP configuration?

I know I have asked  too much help here ( I actualy have done my homework too, but  my network knowledge is too limited), please provide further guidance. I do appreciate your time and efforts to help me for such "complex" problem.

[Hieu Cao]

Just curious, from a workstation on vlan168, can it ping the CAS array at all?

[sean chang]

Thank you,

NO, I can't Ping, it says NOT reachable. but I can ping the individual CAS server.

I watched the Console for a whole, it doesn't show any error log  either, just like nothing happens

[Hieu Cao]

I am currently testing Exchange 2010 in the lab in preparation for the Exchange 2003 migration.  Right now, I've both servers and clients in the same subnet so i don't have the problem with accessng the cas array running in either multicast or unicast mode .  I'll split the servers and client into different subnet/vlan and I will post the result.

[Hieu Cao · Accepted solution]

I had encountered some challenge of getting the clients on diffenet subnet to connect to the CAS array in the native vlan.  I couldn't ping the CAS array until I added the static arp of the NLB mac address and NIC.  Even though I was able to ping the CAS array, OWA did not work when connected to the CAS array, but OWA would work just fine if it's connected directly to the member server of the CAS array.  However, I couldn't assign CAS array's MAC/IP  to a particular port in the switch (I cannot ping the CAS cluster  when it's added) because both of my CAS servers are also virtual machines - Windows 2008 R2 - running on Windows 2008 HyperV R2.

I changed "multicast" to "unicast" and clients from native  and non-native subnets are able to connect to the CAS array just fine.  I will revisit the multicast stuff when I've a chance.

-Hieu

[sean chang]

Thank you , Hieu,  I appreciate your efforts and time spending on this multicast IP routing issue.

It's very interesting to know Unicast works, I can't do anything now becuase our mail system is in production, I'm using another DNS server for Lan168 to point its "cluster IP" to one of the memeber servers, it's working as long as this member server is up.......( or I change DNS to another member should this member goes down).....     it's not ideal but better than nothing..