I saw that a few of the posters are discussing about using LAG. My opinion towards most modest size setup is don't bother. The equipments you are using and the bandwidth you are getting is out cancelling each other in the process.
For LAG, you need multiple clients to spread the bandwidth across multiple physical links. One client is not going to occupy the full bandwidth of the links. LAG is unlike Balance-RR meanwhile only found in unix implementations.
Those benchmark on achieving 100+MBps or even 200MBps are purely sequential access. Once you have more than 1 sequential access to the same volume, the actual bandwidth will quickly degrade to even less than a single 1Gbps bandwidth.
So you can see where the diverge is. You want more bandwidth, you use LAG. LAG need multiple clients access to take advantage of LAG. But as you have multiple clients accessing, the access nature upon the volume turn into random access. Random access are low in performance for spindles.
LAG is only really useful for large SAN with hundreds of spindles that gives enormous IOPs which are in the thousands range. Even for a high performance SAS at 15Krpms, we are only taking about 100+ IOPs.
Take note, SAS interface are duplex while SATA interface are simplex. Hence SATA interfaces just doesn't have the juice to cut for random access. On top of that, unless, you are using the top notch 10Krpms SATA, don't expect good random access.
Furthermore, you are spending on more expensive switches that support LAG.
If you really want network performance from these modest Synology NAS, choose to use iSCSI target with MPIO. I have tested it before using just a single cheap L2 switch with 2 NICs and the performance will easily surpass 100MBps. It's done using Windows Server iSCSI initiator on MPIO. Very easy to setup, the same can be achieved in Unices too
Feel free to read my reports found at
post #27 in this thread
http://forums.hardwarezone.com.sg/storage-clinic-119/need-nas-recommendation-4071349.html