Author

Topic: Is anybody working on a cluster Bitcoin client yet ? (Read 4480 times)

hero member
Activity: 836
Merit: 1030
bits of proof
Sure it is. http://bitsofproof.com.

The site is sparse at the moment but be assured, we are working behind the scenes to fill with content soon.

I am in central Europe too, more specifically in Budapest.

Drop me a mail to [email protected] for personal assistance with the code.


legendary
Activity: 1470
Merit: 1006
Bringing Legendary Har® to you since 1952
Hey, i was just wondering. Is anybody working on a cluster Bitcoin client yet ?
....
Or maybe perhaps it has already been written, but I didn't notice it ?

You are talking about my product the BOP Enterprise Bitcoin Server Smiley

It is modern implementation of a full node that allows running sever instances accessible through an API in form of a load balancing industry standard message Bus.

Product launch is in three days in San Jose.
Interesting. Website links perhaps ?

See you there. I will share more details here thereafter.
Sorry, I am in central Europe currently, no plans of going to the US.
.
.
EDIT:
Is that it ?
http://bitsofproof.com/
hero member
Activity: 836
Merit: 1030
bits of proof
Hey, i was just wondering. Is anybody working on a cluster Bitcoin client yet ?
....
Or maybe perhaps it has already been written, but I didn't notice it ?

You are talking about my product the BOP Enterprise Bitcoin Server Smiley

It is modern implementation of a full node that allows running sever instances accessible through an API in form of a load balancing industry standard message Bus.

Product launch is in three days in San Jose.

See you there. I will share more details here thereafter.
legendary
Activity: 1470
Merit: 1006
Bringing Legendary Har® to you since 1952
So quiet ?

I guess there is nothing like this going on then.

Essentially you have to write the platform for multiple architectures (x86, x64, i686, ARM, etc.) so that you have the ability to work across many systems,

This is a valid scenario, but i was thinking more about scaling the Bitcoin architecture by running multiple identical machines so it could handle tens of thousands or even hundereds of thousands transactions per second.

Certainly such scenario is not possible with the "single-client-doing-eveything" way.
newbie
Activity: 28
Merit: 0
So quiet ?

I guess there is nothing like this going on then.

I don't think many people who want to do this have the ability to. Distributed computing is an amazing thing, and many people would love to do it as far as Bitcoin mining is concerned, but the development process of it is a little tricky.

Essentially you have to write the platform for multiple architectures (x86, x64, i686, ARM, etc.) so that you have the ability to work across many systems, and that's not including writing the binaries for multiple Operating Systems. Then you have to develop a client-server model that specifies how to distribute the work and report back the results, etc., and that just covers the basics of what has to be done.

It takes some work. I wouldn't even bother with a task like that.
full member
Activity: 121
Merit: 103
sounds like you're looking for a variety of HA / distributed architecture features. this would likely be difficult to pull off with bitcoind or any other existing full-node implementation.

this would surely be useful at some later date but doesn't seem that useful atm.

Hey, i was just wondering. Is anybody working on a cluster Bitcoin client yet ?

And by "cluster Bitcoin client" I mean:

- Works using networked sub-clients, each doing only part of the work (transaction signing / relaying / storing blockchain).
- Can be run on multiple machines utilizing all their processors and internet connections
- Can use multiple Internet connections at once
- Can be run on clusters, supercomputers, beowulfs, server farms etc.

Or maybe perhaps it has already been written, but I didn't notice it ?

passing all the structures in question over the network versus doing it all locally / in memory on a single machine could work. however, this will add a lot of latency and likely slow down the operations in question, e.g. fetching tx blocks from the db will take longer, inserting into a remote db will have more latency, individual tx lookups for signature verification will take longer.

btcd can utilize pretty much all the cpu it has on hand, i have seen up to 7 of 8 cores being sucked up for tx signature verification. using multiple network uplinks is more a matter of configuring your network properly than anything that a distributed full-node solution could be expected to do.
legendary
Activity: 1470
Merit: 1006
Bringing Legendary Har® to you since 1952
So quiet ?

I guess there is nothing like this going on then.
legendary
Activity: 1470
Merit: 1006
Bringing Legendary Har® to you since 1952
Hey, i was just wondering. Is anybody working on a cluster Bitcoin client yet ?

And by "cluster Bitcoin client" I mean:

- Works using networked sub-clients, each doing only part of the work (transaction signing / relaying / storing blockchain).
- Can be run on multiple machines utilizing all their processors and internet connections
- Can use multiple Internet connections at once
- Can be run on clusters, supercomputers, beowulfs, server farms etc.

Or maybe perhaps it has already been written, but I didn't notice it ?
Jump to: