Hello,

We're currently trialing MSP360 Ultimate (on Windows Server) attempting to see if it is a potential solution in our use case.

We have to transfer around 70TB from on-prem to S3 weekly (Oracle RMAN backups). They're currently backed up locally using RMAN to an NFS device, and from there we need to get the files up to S3 for longer-term retention. (We've considered just using AWS Storage Gateway for this but we'd like to benefit from the compression of a backup application can offer as every application we've tested gets a solid 3:1 compression on this data).

We have a 5Gbps AWS direct connect, but so far we're only able to push 500 Mbps to 1Gbps from our MSP360 VM.

We're not seeing any obvious bottlenecks on the VM or the on-prem net.

Increasing the chunk size to 256MB and the thread count to 12 and then 24 seems to have marginally improved the speed but it's still far from ideal.

12 CPU VM is around 60% utilization no matter what we have the chunk size or thread counts set to and RAM utilization is acceptable too - around 40-50%.

Can someone share if they know, is there a method to achieve closer to 5Gbps throughput to AWS S3 over an AWS Direct Connect without using the AWS acceleration service? And if so, do you have any suggestions for achieving those speeds with MSP360? Should we separate the backup out into multiple jobs (That's the next thing we're going to try)? I would think that increasing the thread count and chunk size would have the same effect, but perhaps not. Should we run multiple VMs with MSP360 on it, perhaps on different hosts? I wouldn't think this would be necessary since we don't appear to be taxing the VM or the host with just one VM, but again, perhaps someone who knows more could offer some advice here.

Thank you!

Simon