running proxmox on a production critical environment

[amravit]

hello there
i been researching for more then a week now to cover all "what if" scenarios
i couldn't find answers for the following
1- regarding backup & backup performance and speed
we running around 12 servers each one usage estimated between 40 to 150 gb of data
i did read all about vzdump and considering doing weekly backup for all servers on non working hours but i am concern about backup duration and speed .
here is
CPU BOGOMIPS: 89604.35
REGEX/SECOND: 531256
HD SIZE: 36.67 GB (/dev/pve/root)
BUFFERED READS: 60.07 MB/sec
AVERAGE SEEK TIME: 12.90 ms
FSYNCS/SECOND: 1162.51
DNS EXT: 362.10 ms

2- we run around 6 Oracle ebusiness suite estimated between 130 gb to 160 gb per server what is the effect of vzdump snapshot mode on the database
i mean if we restored any vzdump would it be corrupted because of the live database

3- is there any way for incremental backup to solve the issues generated for full backup because of the big space usage

4- host backup recommendation i saw people using rsync rsnapshot what’s your recommendation ?

5- clustering iam planning to add cluster node slave to the master one i have 2 concerns
1- if the master fail the slave is still working correct ?

2- what is the best way to do fail over cluster without san nfs maybe and what’s the correct way to setup it for host machine i have setup webmin on host machine to get the configuration little bit easier

6- any one done fail over cluster for the vps it self ?

7- whats the best way to add more space to host machine once it goe production ( expanding the lvm2 )


sorry for all questions i need some one to think with me loud so nothing goes wrong

regards
amr

 

[tom]

hello there
i been researching for more then a week now to cover all "what if" scenarios
i couldn't find answers for the following
1- regarding backup & backup performance and speed
we running around 12 servers each one usage estimated between 40 to 150 gb of data
i did read all about vzdump and considering doing weekly backup for all servers on non working hours but i am concern about backup duration and speed .
here is
CPU BOGOMIPS: 89604.35
REGEX/SECOND: 531256
HD SIZE: 36.67 GB (/dev/pve/root)
BUFFERED READS: 60.07 MB/sec
AVERAGE SEEK TIME: 12.90 ms
FSYNCS/SECOND: 1162.51
DNS EXT: 362.10 ms

looks like you use slow sata disks. sas disks with 15rpm are much faster but I assume you know this already.

2- we run around 6 Oracle ebusiness suite estimated between 130 gb to 160 gb per server what is the effect of vzdump snapshot mode on the database
i mean if we restored any vzdump would it be corrupted because of the live database

I assume the only way to get reliable backup/restore you need to TESTTESTEST.

3- is there any way for incremental backup to solve the issues generated for full backup because of the big space usage

not with vzdump: in your case I recommend you go for a separate backup solution (backup suite), vzdump is limited in features.

4- host backup recommendation i saw people using rsync rsnapshot what’s your recommendation ?

5- clustering iam planning to add cluster node slave to the master one i have 2 concerns
1- if the master fail the slave is still working correct ?

the vm´s on the node are working, yes. management is only possible on the master so you can bring the master up again or you need to force a node to the new master.

2- what is the best way to do fail over cluster without san nfs maybe and what’s the correct way to setup it for host machine i have setup webmin on host machine to get the configuration little bit easier

6- any one done fail over cluster for the vps it self ?

With Proxmox VE 2.0 (SAN support) you will have a good base to start with this. Some people are already working with DRBD, but this needs deep knowledge and its not yet supported from our team here.

7- whats the best way to add more space to host machine once it goe production ( expanding the lvm2 )

Just use LVM2 commands. But I suggest you do good planning to avoid this to reduce complexity. Again, Proxmox VE 2.0 will give you much more flexiblity in storage. (see roadmap)

sorry for all questions i need some one to think with me loud so nothing goes wrong

regards
amr

I appreciate your questions: for the moment I think your setup is missing the backup suite, maybe you look at Bacula and you test a good mix with Bacula and vzdump - maybe you post in the Bacula forums for a backup startegy - Proxmox VE - vzdump - Bacula?

 

[amravit]

i appreciate your answers and ur time tom
i need your help answering 4 more questions

1- any advice on clustering host machines as fail over active active mode>

2- any advice on clustering vps it self ?

3- any time frame for promox version 2?

4- i have around 8 vlans at work main host machine are hosted on vlan 1
i would like to run 3 diffrent vps on 3 diffrent vlans
please note i have the port of the host machine configured as trunk and i already test it it work with all vlans

the vlan looks like that
1- vlan 1 172.16.1.0 subnet 255.255.255.0
2- vlan 4 172.16.4.0 subnet 255.255.255.0
3- vlan 5 172.16.5.0 subnet 255.255.255.0

 

[tom]

i appreciate your answers and ur time tom
i need your help answering 4 more questions

1- any advice on clustering host machines as fail over active active mode>

2- any advice on clustering vps it self ?

3- any time frame for promox version 2?

1 and 2: not implemented yet on Proxmox VE side. 3 - No release date set yet, we are working on Proxmox VE 2.0.

 

[chopeta]

We use BackupPC for backup/restore here + vzdump backups from PVE web-interface.