Round Robin writing for increased perfomance

+1 vote

Dear MC Team.

Please clarify the usage of clustering nodes according to  https://www.multichain.com/developers/clustering-high-availability/

We use MC as  ledger to store all transactions with consistency. Throughput is very important for us.

In out model, we store no private keys in MC 

1. we prepare transactions with createrawsendfrom 

2. we sign the transaction outside of MC 

3. We submit transaction to MC with sendrawtransaction

Currently we have 2 nodes, and we have plans to increase the amount of active nodes to 6 servers in various locations. No POW, all of them are trusted, and due to that, any of the node can confirm the transaction. 

So, our idea is to submit transactions to the nodes on a round-robin bases, to achieve the maximum transaction/per-second speed. In this scenario, we can simply add mode nodes for higher throughput.

So 20 servers can make ~20 times more transactions per second then.  

Please correct me if I'm wrong ! 

Thank you in advance.

 

asked Jan 5, 2018 by Dilidonka

1 Answer

0 votes
I'm afraid it doesn't work like this, because in a blockchain every node has to process every transaction. You may gain some small performance improvement by transmitting transactions via multiple nodes, but certainly no linear scaling in the number of nodes.
answered Jan 5, 2018 by MultiChain
Ok, I see. Thank you!

But anyway, this will work without any double-spending  and in case if any of the nodes fail, all other nodes will be online.
So we can get a high level fault-tolerance then.
And when the fault node returns back, it will synchronize with others and will get back to normal work.
Right ?
Yes, when the node comes back online it will resynchronize. But about double spending, you need to be careful. Please read the guideilnes on that page about how to about clashing transactions between nodes in the cluster.
...