SyncIQ uses workers on each node, so the more nodes one uses for the transfer, the more workers are involved. SyncIQ will use all the bandwidth you care to provide to it and the workers in each cluster communicate with one another. There is a lot of good documentation on how SyncIQ works available to you.
The number of works that you get is related to the number of nodes in the destination array. The higher the number of simultaneous workers, the more throughput you can get.
There is a configured total number of workers for ALL of the SyncIQ jobs on your cluster and if I remember right, it's set to 80 by default. It's changeable, but you should open a support case and explain what you're trying to do and get some input on whether the source and destinations will take the load.
The majority of the load is on the destination side.
Lastly,if you want higher throughput, get yourself on a new release (at least 7.1) of SyncIQ, especially if you have large files. Older release would tie up a stream for the entire file whereas the newer releases will chunk up the file and distribute it amongst streams. Older releases of OneFS also weren't able to drive a 10Gbps port anywhere near 10Gbps so you need a lot of needs participate in your transfers to get the aggregate you're looking for.
As well as the other replies also consider this: It is possible to run syncIQ job at near wire speed. However what you need to bear in mind is what is doing the work. Firstly you would need a source cluster with enough power to deliver this performance (spindle count). Therefore if you have a 3 X200 cluster you will not achieve this, however with a 4 node X410 depending on the configuration you could drive 1250 MB/s (concurrent). Then you would need the corresponding performing destination cluster. This is again depending on the workflow that the cluster is also supporting at the time with the additional load of the SyncIQ job(s).
With the answer comes a question. What performance do you need to drive for the change rate for replication? Whatever this is the Isilon clusters can be configured to deliver this. Speak to EMC locally and get an Isilon Pre-Sales specialist to help.
We test SyncIQ for our automation product for http://superna.net/eyeglass-isilon/ we have delay simulators and VCE Vbocks for driving data to clusters (ICAP testing or Video streaming). We are capable of doing custom proof of concept if you have a specific latency and RPO and RTO you are targeting (next release of our product will provide trending on RTO RPO based on SyncIQ transfer rates).
We alos have Online free training on the solution in a virtual lab with 2 clusters can be requested here http://superna.net/eyeglassonlineform/
Let us know how we can help
You can create a new smartconnect zone on source and target cluster and add nodes which should be part of synciq and use number of workers to test this out it should work fine.