Network and storage queries in respect to RAC

Question

Network port usage identification in the switch
I’m not sure about the usage of number of sockets/ports in the private switches for Oracle RAC. If we run two node cluster, how the socket usage is like and how many more nodes can be added to the cluster.
For example, if we run two nodes in the cluster, it might use 6 sockets. If the switch is a 24 sockets, I can identify how many more nodes can be added into this cluster. This is to know the scalability of the RAC based on the network component switch.
for information: Our environment has got a gigabit ethernet switch..not sure about the sockets in it.
Storage question
1. How do I identify if a single LUN is shared by multiple databases?
For example, I see only one data volume in ASM, like '+DATA1' which has got many database datafiles, logfile etc. I want to know is there a way I can see a single LUN or a group of LUN been assigned to this volume group in ASM. Is there a recommended practice to do this setup?
I read somewhere "The best practices advise not to mix the LUN with different databases present in the cluster to minimize IO operations." The question 1 is based on this only.
2. If we add/expand the disk or LUN in the volume group, I think the system needs a reboot of all cluster nodes for this to be visible to Oracle ASM. Am I correct? what is restricting not to add dynamically? Storage or OS..no clues for me. Any ideas here...
I have seen the disks which are already part of the disk group can be added dynamically to the ASM disk group but not sure about the addition/expansion of LUN...
Thanks in advance.

Answer 1

Network port usage identification in the switch
I’m not sure about the usage of number of sockets/ports in the private switches for Oracle RAC. If we run two node cluster, how the socket usage is like and how many more nodes can be added to the cluster.About Switch for Private network... you have 2 nodes ,you should 2 ports on switch. If you make bonding on Private interface (server), so 2 nodes .. each of nodes use 2 ports for bonding -> you should use 4 ports on switch.
>
For example, if we run two nodes in the cluster, it might use 6 sockets. If the switch is a 24 sockets, I can identify how many more nodes can be added into this cluster. This is to know the scalability of the RAC based on the network component switch.
for information: Our environment has got a gigabit ethernet switch..not sure about the sockets in it.about port number on switch... you can check from product owner...
Many nodes?... if you have plan to add many nodes... you can buy new switch and then cross to old private switch...
Or example... stack switch feature on Cisco 3750
http://www.cisco.com/en/US/docs/switches/lan/catalyst3750/software/release/12.2_25_see/configuration/guide/swstack.html
Storage question
1. How do I identify if a single LUN is shared by multiple databases?
For example, I see only one data volume in ASM, like '+DATA1' which has got many database datafiles, logfile etc. I want to know is there a way I can see a single LUN or a group of LUN been assigned to this volume group in ASM. Is there a recommended practice to do this setup?You have to know what Storage (EMC/NetApp) you are using? and find document... to help
About ASM:
Example: ASM with Emc .. check on node1
$ export ORACLE_SID=+ASM1
$ sqlplus / as sysdba
SQL> select path, library from v$asm_disk where LABEL='DATA1';
PATH LIBRARY
/dev/raw/raw4 System
that mean we have to check about device name (DATA1) at /etc/sysconfig/rawdevices file
Or
SQL> select path, library from v$asm_disk where LABEL='DATA1';
PATH LIBRARY
ORCL:DATA1 ASM Library - Generic Linux, version 2.0.2 (KABI_V2)
that mean we have used asmlib, So...
# /etc/init.d/oracleasm querydisk DATA1
Disk "DATA" is a valid ASM disk on device [1, 127]
and
# ls -la /dev/* | grep '1\, ' | grep 127
We'll see device name 's used for 'DATA1'
by the way... if you use EMC you can... use "powermt" to check storage on servers
# powermt display dev=all
>
I read somewhere "The best practices advise not to mix the LUN with different databases present in the cluster to minimize IO operations." The question 1 is based on this only.
Mix the LUN with different database may make bottle neck on HW Storage...
2. If we add/expand the disk or LUN in the volume group, I think the system needs a reboot of all cluster nodes for this to be visible to Oracle ASM. Am I correct? what is restricting not to add dynamically? Storage or OS..no clues for me. Any ideas here...On Linux... you have to reboot... to reload library... You can reboot each of nodes ... don't need reboot all nodes
>
I have seen the disks which are already part of the disk group can be added dynamically to the ASM disk group but not sure about the addition/expansion of LUN...You can use "dbca" to help add/expand LUN , after every nodes see new disks.