Memory calculation?

[ejenner]

I'm wondering if my repository server is 'blue-screening' due to a lack of physical memory.

Can anyone help calculate whether or not in my specific scenario we have enough RAM installed?

This is our spec:

- Latest Veeam
- Windows Server 2016 with all updates
- ProLiant DL380 Gen9
- D3600 Drive box with x12 6TB spinners
- Smart Array P441 controller
- 16GB RAM
- RAID 6 55TB volume
- ReFS with 64k allocation unit size
- 10 Gpbs network

Our server has blue-screened 4 times in the last month and we're struggling to work out why. Is it possible to underspecify physical memory when you have a setup like this or should Windows be able to keep the situation under control and keep the server running even if you don't have quite enough RAM? I would naturally assume it could handle it but because we're scratching our heads we're having to think outside the box.

[DGrinev]

Hey,

I don't think that is caused by the lack of RAM, but anyway you can check the requirements and limitations in the User Guide here:
1. Task Limitation for Backup Repositories.
2. Backup Repository Server system requirements.
Please share the number of concurrent tasks in the repository settings and how many jobs concurrently initiated at the peak time. Thanks!

[PTide]

my 2 cents:

16 GB of RAM does not seems to be enough for ReFS - 1Gb per TB of space is reported to be the minimum recommended amount.

Thanks

[ejenner]

It would be really nice to know how to properly calculate it. Especially if we have to justify the extra purchase, which I'd imagine most of us would have to do. As we don't know that our current amount of RAM is not enough and we can't say what would be... that's a difficult case to make.

[ejenner]

Hey,

I don't think that is caused by the lack of RAM, but anyway you can check the requirements and limitations in the User Guide here:
1. Task Limitation for Backup Repositories.
2. Backup Repository Server system requirements.
Please share the number of concurrent tasks in the repository settings and how many jobs concurrently initiated at the peak time. Thanks!

Our system is pretty lightly loaded at this point as we're only just starting our implementation. We have 21 VAW jobs configured at the moment. They're set to kick off in three tranches. Divided by 3. So the first 7 jobs will kick off at 2am, then the next 7, then the final 7.

[foggy]

You can calculate based on the system requirements referenced above. But it is already clear than 16GB is not enough for that amount of concurrent jobs, since every single one requires up to 4GB.

[ejenner]

The parallel processing settings should cap it at 2 concurrent jobs by default? I'm not sure?

I also saw: "It is recommended to configure 2 GB RAM per core"

Here: https://helpcenter.veeam.com/docs/backu ... tml?ver=95

[foggy]

The parallel processing settings should cap it at 2 concurrent jobs by default? I'm not sure?

If the max concurrent tasks number is set appropriately, yes.

I also saw: "It is recommended to configure 2 GB RAM per core"

It's up to 4 GB RAM for 64-bit OS.

[ejenner]

I think with our spec we'd be looking at a minimum of 64GB to meet the recommendation for the ReFS based calculation. Our 8 core processor would require 32GB.