I recently deployed a  new 32TB Data Domain DD3300 system. The initial configuration is easy and familiar enough. Connect to the system via the serial cable, setup iDRAC, and run the initial configuration wizard. Afterwards the rest of the configuration can be performed via SSH and/or the GUI.

While continuing with the configuration, I noticed I could not create an aggregate Ethernet interface. No LACP or Etherchannel! So what if my Ethernet interface goes down, for whatever reason?

Aggregate or virtual interface

To recap, in “normal” Data Domain systems you’re able to create a virtual interface. You’ll group physical Ethernet interfaces under this virtual interface, and select an aggregation protocol. Typically this is LACP.

The DD3300 does not support this option, as you can see in this screenshot:

It took me a while to figure this out, until I ran into KB article 518492. Apparently, only the DD3300 does not support LACP. All other new models do. I’m going on a limb here, but I suspect it might have something to do with the fact the DD3300 is really a Dell server running a VMware hypervisor and a Data Domain Virtual Edition VM. It’s kinda hard to have a VM do LACP to a VMware vSwitch…

This introduced two problems. First, we need redundancy from an Ethernet perspective. If the ethMa link goes down for whatever reason (e.g. switch failure), this would mean we’d lose connectivity to the system over that assigned IP.

It also means that we’d only have 1GbE of bandwidth behind a single IP address. With LACP you’re able to aggregating multiple 1GbE links into one virtual interface, hence you could set up multiple 1GbE streams. While not getting the full theoretical bandwidth (LACP is designed for redundancy, not performance), you’d at least get more than 1Gbit…

DDBoost Interface Groups to the rescue

A bit of further research pointed us in the direction of DDBoost interface groups. With this approach, you assign unique IP addresses to a number of interfaces. In our case, we assigned a second IP address to interface ethMb. You then group these two IP addresses into a DDBoost interface group.

As a result, the DDBoost protocol and the Data Domain will load balance the connections across the available IP addresses, and thus the available physical interfaces.

Configuration is simple: you can add the IP addresses to the DDBoost interface group via the GUI…

Or you add them via the CLI, using the command:

ifgroup add default interface <ip address>

Repeat this command for the additional IP addresses. You can only add IP addresses of different physical interfaces.

My thoughts on the lack of LACP in the DD3300

I must admit I was a bit surprised that the DD3300 does not support LACP. The Data Domain spec sheet does not explicitly show this, and it took a bit of digging to find a KB article that explained the difference. Since this customer has a global standard configuration for Data Domains, the lack of LACP means we will deviate from this standard. This could be confusing for the people that need to keep these systems running, since we’ll now have two different active configurations.

Fortunately, the DDBoost interface groups still enable us to provide Ethernet redundancy and to utilize more than the bandwidth of a single interface. According to Dell EMC, interface groups are preferred over LACP anyway. It will provide greater bandwidth, since LACP is not optimized for performance. It’s easier to configure, without disruption to the link and no additional switch configuration.

That being said: DDBoost interface groups only apply to the DDBoost protocol (it’s kind of obvious in the name). If you are accessing your Data Domain 3300 over NFS or SMB (not CIFS, fix the GUI!), you cannot use these interface groups. So in these cases, you would still utilize one IP and thus, one port. Be aware!