Okay, thanks for your explanation. So basically there is no much benefit having the 4x100Gbits and hpc environment for secondary storage traffic when using a separate nic.
How would you size the nic for secondary when using 10ssds per host? Can you give me any number -----Original Message----- From: Jayanth Babu A <[email protected]> Sent: Freitag, 29. März 2024 12:21 To: [email protected] Subject: Re: Understanding Secondary Storage Traffic Alright. CephFS and RBD stay in different Ceph pools. To clarify, this is how CloudStack does it 1. Upload of Guest Images and ISOs: The secondary storage VM mounts the NFS share (from CephFS) and writes the images to CephFS pool. 2. Instance Launch for the first time: The RBD image gets created from the QCOW2 (or compatible) stored in NFS (CephFS). This is performed by the hypervisor which mounts the NFS share and creates a RBD image. 3. Disk Snapshots: Hypervisor mounts the NFS share (also does the format conversion from RAW to QCOW2) and stores it on NFS share (CephFS). Now coming to your question, there is no inter-pool copy in Ceph for any transactions, so consider some additional copying over secondary storage and hypervisor plus protocol translations between RBD and NFS. A secondary storage instance is bound on the “management” network of hypervisors so can make use of your 4 * 100 Gbps in presence of a trunk for management and other networks. In case you’re planning for external NFS server by any chance, please plan your network interface of the NFS server in such a way that the secondary storage and hypervisors are able to serve all the 3 points mentioned above at the scale of the infrastructure. Regards, Jayanth Reddy From: R A <[email protected]> Date: Friday, 29 March 2024 at 4:06 PM To: [email protected] <[email protected]> Subject: RE: Understanding Secondary Storage Traffic Yes, CephFS over NFS (ganesha) From: R A <[email protected]<mailto:[email protected]>> Date: Friday, 29 March 2024 at 2:10 PM To: [email protected]<mailto:[email protected]> <[email protected]<mailto:[email protected]>> Subject: Understanding Secondary Storage Traffic Hello Community, i am planning a cloudstack hyperconverged cluster with Ceph as primary and secondary storage. I already know that snapshots, volume backups and images are transferred via secondary storage interface, but I have problems in understanding the correct flow. Maybe someone can explain it to me , because i need to understand which loads are transferred via secondary to size the interface correctly. In my setup I ve 2x nics with each 2x 100Gbits ports for eph cluster network and public network. So in sum Ceph has 400Gbits on each host. So now the documentation says that cloudstack will transfer snapshots/images/backups via secondary storage. But I don't get it because the data is hosted by ceph and ceph has a capability of 400Gbit. So when I for example use 30Gbits for secondary storage will this limitate snapshots/images/backups to 30Gbits? Or does it mean that cloudstack use the secondary storage network for management things, but the data are still transferred by ceph? BR Reza Disclaimer *** This e-mail contains PRIVILEGED AND CONFIDENTIAL INFORMATION intended solely for the use of the addressee(s). If you are not the intended recipient, please notify the sender by e-mail and delete the original message. Further, you are not authorised to copy, disclose, or distribute this e-mail or its contents to any other person and any such actions are unlawful and strictly prohibited. This e-mail may contain viruses. NxtGen Datacenter & Cloud Technologies Private Ltd (“NxtGen”) has taken every reasonable precaution to minimize this risk but is not liable for any damage you may sustain as a result of any virus in this e-mail. You should carry out your own virus checks before opening the e-mail or attachment. NxtGen reserves the right to monitor and review the content of all messages sent to or from this e-mail address. Messages sent to or from this e-mail address may be stored on the NxtGen e-mail system. *** End of Disclaimer ***NXTGEN***
