File serving interfaces
NAS servers provide the following file serving physical interfaces:
Models 5200 / 5300
- tg1 - tg6 - 10 GbE interfaces (SFP+)
NAS module
- tg1 - tg6 - 10 GbE interfaces (SFP+)
Models 4060 / 4080 / 4100
- tg1 - tg4 - 10 GbE interfaces (SFP+)
Model 4040 and earlier
- ge1 - ge6 - 1 GbE interfaces (RJ45)
- tg1 - tg2 - 10 GbE interfaces (XFP)
File serving physical interfaces enable network clients to access an EVS on the storage server. These interfaces are commonly added together in a link aggregation to increase redundancy and throughput of data.
Link aggregation
In a link aggregation, one or more file serving interfaces are grouped to form a single logical interface. This functionality can increase bandwidth capability and create resilient and redundant links. Aggregating multiple network links does not increase performance of a single client TCP connection, but it does enable more individual connections to be served faster by using more available links and by reducing contention within a link. An aggregation also provides load balancing where the processing and communications activity is distributed across several links in a trunk. Therefore, aggregations provide higher link availability and increased Link Aggregation Group (LAG) capacity.
An aggregation is assigned a unique MAC address which is different on each cluster node. Each aggregation can have multiple IP addresses. It is possible to configure an aggregation without any IP addresses, but this prevents communication through that interface. For example, in a cluster, an aggregation associated with an EVS appears on all nodes but is only active on the node that the EVS is running on because the EVS holds the IP address. If the EVS fails over onto another node, the IP address moves with the EVS, activating the aggregation on the new node.
The server supports static aggregations. It also supports the Link Aggregation Control Protocol (LACP) for dynamic aggregations.
To view the status of an aggregation, navigate to the Link Aggregation page as shown below:
Link Aggregation Control Protocol (LACP)
The server supports the Link Aggregation Control Protocol (LACP), which it uses to manage an individual link's transmission state (within a Link Aggregation Group). The server controls the LACP relationship between multiple switches. The server determines which network interfaces are in use and can bring up alternative network interfaces during a failure. For example, if the server does not receive any LACP messages from the primary switch (the waiting time is determined by the configured LACP timeout), the server can use the network interfaces connected to the secondary switch instead.
LACP aggregates are not automatically created or populated. The administrator must first create an aggregate interface, then enable LACP on that interface.
LACP timeouts
The server supports both short (one second) and long (30 second) LACP timers. A short timeout is three seconds (three x one second). A long timeout is 90s (three x 30 seconds). Therefore, the link times out after three missed messages. Long timeouts are recommended to upgrade upstream network devices without causing path failover on the server. The default NAS setting is a short timeout.
Typical LACP configurations
There are several typical configurations when using LACP with NAS servers and multiple switches for resiliency:
- Split-LAG with Layer-2 redundancy
- Split-LAG with Layer-2 redundancy and increased bandwidth
- Split-LAG with a single logical switch
Split-LAG with Layer-2 redundancy ( NAS module )
With LACP enabled, the two NAS modules in a Unified setup can be connected to a pair of independent switches configured with one link aggregation (over two file-serving interfaces), as shown in the example below. In this setup, only one of the ports in the aggregate will be active; since the switches are independent from each other LACP will disable the other interface, as illustrated by the dotted lines.
Split-LAG with Layer-2 redundancy (HNAS cluster)
With LACP enabled, a pair of clustered HNAS servers can be connected to a pair of independent switches configured with one link aggregation (over two file-serving interfaces), as shown in the example below. In this setup, only one of the ports in the aggregate will be active; since the switches are independent from each other LACP will disable the other interface, as illustrated by the dotted lines.
Split-LAG with Layer-2 redundancy and increased bandwidth
This configuration is a continuation of the previous example, using a link aggregation over four file-serving interfaces to two separate switches. This doubles the active and standby links, as shown in the example below:
Single logical switch with Layer-2 redundancy and increased bandwidth
This configuration includes a pair or switches connected in such a way as to appear as one logical switch. The HNAS servers are configured with a link aggregation over four file-serving interfaces, as shown in the example below. In this scenario since the switches are logically connected and no longer independent, a single port channel will extend across the switches and use all available links, and therefore all the ports will become active. This is the most optimal configuration for link aggregation across two switches.
VLAN interfaces
A physical network can be partitioned into multiple, isolated distinct broadcast domains called virtual LANs or VLANs. The NAS server supports a maximum of 256 VLANs with a range of VLAN IDs from 1 to 4094 (0 and 4095 are invalid).
A server can provide access to a VLAN using a VLAN interface on an aggregate interface. Administrators can create a VLAN interface for each tagged VLAN for each aggregate interface over which the NAS server needs to communicate. For example, VLAN 1 on ag1 is different from VLAN 1 on ag2.