Dual gb pci-e x1

[Sjakie]

G'day [H]

Network and server overhaul is around the corner, just planning the changes that will fit my budget for now.
At the moment my esx box is taking all the hits as a test mule, storage, webserver and everything you can imagine running on it. Far from ideal which made me rethink my storage.
I currently have hardware capable of running my mdadm array with the possibility to expand with a few disks as soon as them prices start to sink but there's a problem. No unused pci-e slots except for a x1 v2.0. I planned on using a dual nic with tcp offloading and possibily lacp with RR to get the most out of it but all the nics are x4. I know the bandwidth of a x1 lane is limited to 5gbps (?) for pcie 2.0 and the more recent intel nics are in fact pcie 2.0 compatibly.

All dual (and up, doh) are advertised as x4, x8 and x16 compatible but shouldnt a dual gbit nic work just fine in a pcie x1 2.0 slot? bandwidth shouldn't be the problem, even in a full duplex scenario which is highly unlikely with storage.
I would love to try out an intel ET series on my storage box and enjoy 2gbit transfers as it is more in line with the capabilities of todays disks and raid arrays. On the other hand, i don't want to invest in an ET series with tcp checksum offloading if it isn't going to work in my storage box as all the other servers and clients have cpu's idling most of the time anyway.

So, what's your input on this?

New here by the way, just the kind of forum I was looking for all this time

 

[dashpuppy]

G'day [H]

Network and server overhaul is around the corner, just planning the changes that will fit my budget for now.
At the moment my esx box is taking all the hits as a test mule, storage, webserver and everything you can imagine running on it. Far from ideal which made me rethink my storage.
I currently have hardware capable of running my mdadm array with the possibility to expand with a few disks as soon as them prices start to sink but there's a problem. No unused pci-e slots except for a x1 v2.0. I planned on using a dual nic with tcp offloading and possibily lacp with RR to get the most out of it but all the nics are x4. I know the bandwidth of a x1 lane is limited to 5gbps (?) for pcie 2.0 and the more recent intel nics are in fact pcie 2.0 compatibly.

All dual (and up, doh) are advertised as x4, x8 and x16 compatible but shouldnt a dual gbit nic work just fine in a pcie x1 2.0 slot? bandwidth shouldn't be the problem, even in a full duplex scenario which is highly unlikely with storage.
I would love to try out an intel ET series on my storage box and enjoy 2gbit transfers as it is more in line with the capabilities of todays disks and raid arrays. On the other hand, i don't want to invest in an ET series with tcp checksum offloading if it isn't going to work in my storage box as all the other servers and clients have cpu's idling most of the time anyway.

So, what's your input on this?

New here by the way, just the kind of forum I was looking for all this time

Just because you have 2 1gb nic's and plugged in doesn't mean you will get 2gb link, you need to team them. However to team them you need the hardware & switch, and im pretty sure * i saw on here * that you need more than 1 connection through it to affectively work.

 

[Sjakie]

I am aware of the network requirements and limitations of nic teaming, trust me I didnt expect it to just work . With round robin for instance its possible to load balance iscsi traffic between esx and storage, on a dedicated vlan ofcourse.
It's a poormens solution in many ways but 10gbit ethernet is way out of my budgte .

 

[dashpuppy]

I am aware of the network requirements and limitations of nic teaming, trust me I didnt expect it to just work . With round robin for instance its possible to load balance iscsi traffic between esx and storage, on a dedicated vlan ofcourse.
It's a poormens solution in many ways but 10gbit ethernet is way out of my budgte .

im interested to see this then i run iscsi & managed vlans but iscsi is running with wss2008..

Dash.