DSM offers four ways to join two LAN ports into one bond, and they are not four speeds of the same thing. Two need a cooperating switch, two do not, and one of them is not about speed at all. Pick by what your switch can do and what you actually want.
The Create Bond dialog in DSM shows a short list of modes with short names and even shorter descriptions. Choosing wrong does not explode anything — it quietly gives you a bond that does less than you think, or one that spends its life fighting your switch. Here is each mode in plain words, with its one honest job.
DSM balances traffic sent and received by the NAS across both ports entirely by itself. It works with any switch — even two different switches — because the switch is never told anything special.
One warning straight from DSM's own help: with this mode, do not enable link aggregation on the switch, even if the switch supports it. The two ends would both try to manage the same pair of links and end up fighting. For a home or small office with an ordinary switch, this is the default choice.
This is the standard, negotiated form of aggregation. The switch must have LACP enabled on the exact ports the NAS is plugged into (usually as a LAG group), and both ends then agree to treat the pair as one logical link.
If you use more than one switch, they must be stacked into one logical unit — a plain second switch will not do. When the managed switch exists, this is the cleanest-behaving mode: the switch genuinely sees one fat link, and failover between members is handled by the protocol itself.
Balance XOR spreads connections across the links by hashing addresses, and it expects static link aggregation configured on the switch to match. There is no negotiation — both sides are simply told "these ports are a pair" and trusted to agree.
That trust is its weakness. DSM's help notes that if the switch does not support aggregation, outgoing traffic can still be balanced, but a lost link can be detected without revealing the cause. Mismatched static setups degrade quietly instead of failing loudly. Today, LACP or Adaptive Load Balancing cover almost everyone better.
Only one port carries traffic. The other sits idle, watching. When the active interface goes down — cable pulled, port dead, switch rebooting — a standby port takes over and the NAS stays reachable on the same address.
It offers zero extra bandwidth and needs nothing from the switch, and for a NAS whose real job is simply being there — backup target, file server, home services — it is honestly the best mode of the four. Not every upgrade is about speed.
Answer three questions. Is the switch managed, with LACP available? Choose IEEE 802.3ad. No managed switch, but several users at once? Choose Adaptive Load Balancing. Is uptime worth more than throughput? Choose Active/Standby.
And Balance XOR? Only when the switch offers static aggregation but not LACP — a shrinking corner case. Picking the mode is a two-minute decision once the questions are honest; picking it by which name sounds fastest is how the trouble starts.
Watch the video for the card and addressing setup, then come back to this mode table before you click Create Bond.
▶️ Watch Full VideoWith Adaptive Load Balancing or Active/Standby, yes — that is one of their advantages. LACP and Balance XOR expect both links to end on the same switch, or on switches stacked into one logical unit.
The bond is a new interface with its own settings. Give it the NAS's usual static IP and remove that address from the individual member ports, so the network keeps finding the NAS exactly where it always was.
It appears when Open vSwitch is enabled, which Virtual Machine Manager needs. It balances traffic without special switch support and can span two switches. If you are not running virtual machines on the NAS itself, the standard four modes are the ones to compare.