This topic has been marked solved and closed to new posts due to inactivity. We hope you'll join the conversation by posting to an open topic or starting a new one.
I have a ReadyNAS 626X using one of the 1Gbps network interfaces (connected to a 1Gbps switch). I have about 60-70 users, though only a proportion of those will ever be accessing it at the same time.
If I copy to or copy from the NAS, I am getting about 110-120Mbps. I've noticed that there is a 10Gbps NIC on the NAS. Is there any reason that I shouldn't use this instead? Will I get anything from it when still connected to the 1Gbps switch, or is my copy speed a resource issue with the load on the NAS?
I've determined that the slowdown was being caused by traffic passing from subnet to subnet through the firewall. The 10Gbps NIC would never have made a difference.
Thanks for your suggestions though, I will take a look at those in time.
If I copy to or copy from the NAS, I am getting about 110-120Mbps.
That is expected - a gigabit connection has a max throughput of about 120 MiB/sec (1 MiB being 8,388,608 bits). And that doesn't include packet overhead or account for latency. So in practice you'll generally somewhat less on sustained large file transfers.
I've noticed that there is a 10Gbps NIC on the NAS. Is there any reason that I shouldn't use this instead? Will I get anything from it when still connected to the 1Gbps switch.
It won't go any faster than 1 Gbps when you connect it to your switch. So there is no benefit in simply switching to it with your current network.
You might benefit from bonding. What switch are you using (manufacturer and model)? Are you using multiple switches for your users?
Of course upgrading the switch infrastructure to support 10 gbps between the switches and also to provide a 10GbaseT port for the NAS would be ideal - but that could be expensive.
Thanks for your reply - however, just to clarify, I'm getting 110-120Mbps (so about 15MB/s) - way under the capacity of a 1Gbps connection. This is why I was hoping that switching to the 10Gbps NIC might help, assuming that the bottleneck is the traffic on the existing NIC, and not the NAS's resources.
Thanks for your reply - however, just to clarify, I'm getting 110-120Mbps (so about 15MB/s) - way under the capacity of a 1Gbps connection.
How are you measuring this? Is the PC connected using gigabit ethernet? Also, is this when the system is under full load (60-70 users)?
If you are using a managed switch, then you could implement bonding (using two network connections). LACP for instance should work well for your usage. What switch are you using?
Yes, the PC I'm testing with has a 1Gbps NIC. I'm measuring just using the speed reported by the Windows file copy process.
It's difficult to know how many people are actively using the NAS at any given time. Certainly not the office capacity of 60-70 users. I could further test this by running the file copy out of hours.
The switches are manageable, but we use them dumb with the network configuration all on the firewall. Is it fair to say that there's no downside to switching to the 10Gbps NIC? It's either going to help, or just operate at the existing speed anyway?
The switches are manageable, but we use them dumb with the network configuration all on the firewall. Is it fair to say that there's no downside to switching to the 10Gbps NIC? It's either going to help, or just operate at the existing speed anyway?
If you are connecting to a gigabit switch, then there is no benefit at all. It shouldn't hurt either.
However, configuring bonding on the switch and the NAS could provide a noticeable performance boost for your users. In this mode you could bond all four NICs together, and quadruple the NAS network bandwidth.
Options on bonding are to
use round-robin on the NAS and static LAG on the switch
use LACP on both the NAS and the switch
LACP is the better option for you if the switch supports it.
However, if your switches are connected to each other with gigabit ethernet, you'd also want to create bonded connections between the switches.
I've determined that the slowdown was being caused by traffic passing from subnet to subnet through the firewall. The 10Gbps NIC would never have made a difference.
Thanks for your suggestions though, I will take a look at those in time.
