General questions about Nakivo Backup

[Jorge Martin San Jose]

Hi all

We are thinking of changing our current Microsoft DPM backup system for Nakivo and I have many doubts about the storage configuration and the technology used by nakivo.

the idea is to have a server with ssd disks with a storage of 40TB to make copies to disk, all the disks will be SSD Enterprise.

To save space, we are interested in making full syntentic copies of our VMs, at this point is where my doubts begin after the experience with Microsoft DPM

- What file system does Nakivo use for full synthetic backups?

- It is necessary to have ReFS to be able to perform full synthetic backups? (Do you use nakivo refs for block cloning?)

- There is some guidance for the implementation of the file system underlying the Nakivo repository. I mean things like interleave size, block size, file system. Our idea is to set up the server with Windows using Storage Spaces to be able to expand the pool if needed.

 

My experience indicates that without an efficient storage configuration, random write backup speed drops dramatically.
The Competitor forum is full of posts with performance issues and the cause is undoubtedly poorly configured storage.

Refs uses a large amount of memory for the deletion of large files such as a purge of old backups. I have seen it use up to 80GB of RAM just to perform the purge and this is due to storage performance.

ReFs uncontrolled metadata restructuring tasks and cache utilization also bring down performance.

 

Thank you very much for clarifying my initial doubts, surely after receiving your answer I will have new questions.

 

Regards

Jorge Martin

 

 

 

 

[The Official Moderator]

On 11/25/2023 at 8:30 AM, Jorge Martin San Jose said:

Hi all

We are thinking of changing our current Microsoft DPM backup system for Nakivo and I have many doubts about the storage configuration and the technology used by nakivo.

the idea is to have a server with ssd disks with a storage of 40TB to make copies to disk, all the disks will be SSD Enterprise.

To save space, we are interested in making full syntentic copies of our VMs, at this point is where my doubts begin after the experience with Microsoft DPM

- What file system does Nakivo use for full synthetic backups?

- It is necessary to have ReFS to be able to perform full synthetic backups? (Do you use nakivo refs for block cloning?)

- There is some guidance for the implementation of the file system underlying the Nakivo repository. I mean things like interleave size, block size, file system. Our idea is to set up the server with Windows using Storage Spaces to be able to expand the pool if needed.

 

My experience indicates that without an efficient storage configuration, random write backup speed drops dramatically.
The Competitor forum is full of posts with performance issues and the cause is undoubtedly poorly configured storage.

Refs uses a large amount of memory for the deletion of large files such as a purge of old backups. I have seen it use up to 80GB of RAM just to perform the purge and this is due to storage performance.

ReFs uncontrolled metadata restructuring tasks and cache utilization also bring down performance.

 

Thank you very much for clarifying my initial doubts, surely after receiving your answer I will have new questions.

 

Regards

Jorge Martin

 

 

 

 

@Jorge Martin San Jose , thank you for reaching out to us. In the current NAKIVO implementation, we support two types of repositories: forever incremental and incremental with full.

The incremental with full repository type requires more space due to separate full recovery points but generally exhibits better reliability during emergencies such as power outages or hardware failures.

The forever incremental repository has a different structure and may require occasional maintenance, similar to a "defragmentation" process. Both repository types are compatible with the ReFS filesystem, although we are currently investigating potential performance differences compared to other file systems like NTFS or FAT.

For larger repository sizes, especially those exceeding 10 TB, we recommend using the "Incremental with full repository" type. However, it's still feasible to create multiple repositories, each ranging from 8-10 TB, in the described scenario. If you opt for the Incremental with full repository type, we suggest using Active full backups instead of Synthetic full backups.

Although the final backup size remains the same for both types, they differ in how a full backup is created. Synthetic full backups leverage previous backup information, requiring consistency checks, while active full backups pull data directly from the VM. Feel free to explore these options and test the described layout, different repository types, and overall NAKIVO performance using our Free trial version.

Should you require further assistance, please don't hesitate to reach out us on the Forum or using this email: support@nakivo.com.