How to avoid unnecessary use of network backing up VMs to a NFS storage

[fbarrio]

Alright, let me explain my scenario

I have a couple of Esxi hosts with 6 VMs on each

I have 2 NFS Linux Storages ( storage1 an storage2) and nakivo running in one of those storages (storage2) locally

 

Right now, if I want to backup a VM, it goes something like this:

VMs get snapshotted, and new files are created on their respective storages

Nakivo starts copying the files to the repository, BUT, its goes like this

storage1 -> esxi host -> nakivo repository (storage2)

storage2 -> esxi host -> nakivo repository (storage2)

Instead of going directly like this

storage1 -> nakivo repository

storage2 -> local nakivo repository  (without using the network at all)

So, right now, nakivo is creating backups at around 50-60MB/sec but they are using all the bandwidth available on the esxi hosts, completely wrecking their performnce while backups are running.

wouldn't make sense for nakivo to pull the snapshot files directly from the esxi storages? is that posible?  maybe adding the storages directly on the nakivo admin panel and indicating were each VM is stored?

 

 

[Official Moderator]

In this case we would recommend you the following NAKIVO layout:
1. Deploy NAKIVO transporter on both ESXI hosts to be able to use HotAdd datatransfer mode.
2. Install NAKIVO transporter on NFS storage, in case it is supported.
In this case both involved transporters will perform direct read and write operations on the source and target sides correspondingly and also have direct connection between each other.

For detailed analysis, please send a support bundle (http://www.nakivo.com/helpcenter/display/NH/Support+Bundles) and the following information to support@nakivo.com :
- in which way ESXi storages are connected to the corresponding ESXi hosts
- which hosts are used as NFS repositories

[fbarrio]

ok, i deployed the transpoters but i see no change..

Nakivo is installed locally on a storage1, that provides datastore via NFS to the 2 esxi hosts

i also have another storage2 that provides datastore via NFS to the 2 esxi hosts, but for this test, im not using it.

I have VMs on esxi1, with datastore1, and when i backup them, they go at very low speed, and use the network

all nakivo has to do is copy the snapshots locally, as the datastore and nakivo are the same server..

what am i missing?

[Official Moderator]

We need to analyze your log files to investigate why you encountered this issue. Please generate and send a support bundle https://helpcenter.nakivo.com/display/NH/Support+Bundles to support@nakivo.com so our 2nd Level Support team can further investigate the issue you've experienced and suggest a proper solution.

Also, the networking diagram will be really useful.