Forum
Welcome, Guest
Username: Password: Remember me

TOPIC:

Volume size limitation 5 years 1 month ago #1791

  • Giorgio Catena
  • Giorgio Catena's Avatar Topic Author
  • Offline
  • Posts: 30
Hi,
considering the discussion made few months ago: "HA-Lizard - NoSan 2 Node configuration" I'm guessing if the sync has some limits on the volume size.
I'm going on with the project and my idea is to link 2 HPDL380G10 with 12TB of RAID volumes to sync. The two nodes should mounts 3 machines one with a high write ratio (and a big volume to store images coming from a tvcc system) one, acting as a manager, with a low write ratio (the manager system) and a monitoring system (centreon). The idea is to place the big and disk consuming machine on the main node and the other two machines on the secondary nodes. The link between the two machines would be on a teamed gigabit or, better, on a teamed 10GB card.

Would it be possible or do you know if there could be any kind of limitation?

Please Log in or Create an account to join the conversation.

Volume size limitation 5 years 1 month ago #1792

  • Salvatore Costantino
  • Salvatore Costantino's Avatar
  • Offline
  • Posts: 722
Your only limitation would be what XenServer supports.. With regards to HA-Lizard/iSCSI-HA. We use DRBD for replication and 12TB of storage will not be a problem for that, nor do I expect there to be a problem with XenServer iSCSI-SR (which is how we expose the storage).
You should check, just to be sure, what the max disk size is for a single VM in XenServer

Regarding your deployment plan. You are correct to run your heaviest workload on the primary server as that will have direct access to the storage and perform as if it were DAS (very fast). VMs running on the slave host will need to traverse the replication link to get to the storage, but, if they have low write requirements, they would operate fine.

Please Log in or Create an account to join the conversation.

Volume size limitation 5 years 1 month ago #1793

  • Giorgio Catena
  • Giorgio Catena's Avatar Topic Author
  • Offline
  • Posts: 30
Thanks for the fast answer Salvatore,
the machine will have a system disk plus an archive disk for tvcc streams. This is why also I'm guessing if to opt for a raid 5 / 6 or a raid 1+0 configuration in order to support the data rate (if i remember properly the master would have a 30% of impact in performance against the 70% impact on the slave. If I adopt a raid configuration that multiply the disk performance on read-write (Raid1+0) I should balance this properly.
The max VM volume size would be around 10 TB.

Please Log in or Create an account to join the conversation.

Volume size limitation 5 years 1 month ago #1794

  • Giorgio Catena
  • Giorgio Catena's Avatar Topic Author
  • Offline
  • Posts: 30
One other thing.
Do you think, in case a problem to provide a 10 GB connection that a dedicated 2 x 1GB connection would be sufficient for the drdb sync of such a big volume?

Please Log in or Create an account to join the conversation.

Volume size limitation 5 years 1 month ago #1795

  • Salvatore Costantino
  • Salvatore Costantino's Avatar
  • Offline
  • Posts: 722
DRBD initial sync would take a long time.. maybe a day or so, but only occurs one time when the storage is first initiated.
Regarding 10G vs 1G. It really depends on the application, disk performance, etc.. If your application is diskbound, then consider 10G for replication which would eliminate any potential bottleneck for VMs running on the secondary host.

Please Log in or Create an account to join the conversation.

Volume size limitation 5 years 1 month ago #1796

  • Giorgio Catena
  • Giorgio Catena's Avatar Topic Author
  • Offline
  • Posts: 30
Thanks for the answer, I'm trying to insist to provide a double 10 GB direct link in order to create a correct replication link. Regarding the witness IP and the split brain risk, one IP will be the gateway that is the main gateway of the company that should always be reacheable but i will add some other else in order to have a most safe check in case of a reboot of the gateway that could always be done.

Please Log in or Create an account to join the conversation.