Scalability

[jib]

I think a lightweight client would need to trust the fast node it talks to that sees every transaction.

[TopSoil]

That's not bad.  Actual network bandwidth will be higher (the way the network is connected you get the same transaction multiple times from your peers).  You won't be running an always-connected-network node on your iPhone, but any low-cost server will give you twenty times that bandwidth per month.  And 18GB isn't much disk space in these days of terabyte hard drives.

Not bad? huh? That seems totally broken to me.
I'm less worried about storage space than the bandwidth and time it will take a new node to enter the network.
Like you said the blocks are duplicated so a given node must download some multiple of the transactions a day. That is a tremendous amount of downloading.
Also any new node must download 216GB for each year the network has been this large!

Doesn't the system also just completely break at some point when the rate of transactions is too high? I could be wrong since I'm not sure exactly how it is decided which transactions go into a particular block. Again the pdf is kind of lacking.

Also isn't there a problem when 2 computers both solve a block before they realize the other one has? This will become more and more likely as there are more nodes.

Eventually, if Bitcoin survives and gets as popular as credit cards for paying for stuff I expect somebody will create a compatible version with a more efficient network structure (maybe by that time there will be

But it seems like a lot of these things aren't just issues of network structure. They seem inherent in the cryptographic design so it wouldn't be that straight forward to replace. Plus why not address these issues now if only to instill faith that the system is well thought out rather than just wave hands and say "computers will be faster" or "we'll tack on some better networking later"


Again I think the general idea is great and I hope the developers have good answers for all these questions. I just want to understand it/make sure it has been thought through carefully enough. I sincerely hope that it actually can work. I'd love to never have to use paypal again.

[Gavin Andresen]

And I expect most of us will be running lightweight clients that just keep our wallets, sign transactions, and send and receive transactions to the ultra-fast nodes that ARE looking at every transaction.

Is this possible? What would this look like? From a technical perspective what does a "lightweight client" look like for you? My understanding is that the Bitcoin client needs the entire block chain in order to establish trust.

I'm imagining:

A lightweight client would have a wallet with coins in it (public+private key pairs).

And a secure way of sending messages to, and getting messages from, any of the ultra-fast, always-connected heavyweight nodes.

The lightweight client sends money by:
  creating a transaction (signing coins with the private key)
  sending the signed transaction securely to the ultra-fast server, which puts it on the network.
  receiving confirmation that the transaction was valid and sent, and updating its wallet (marks coins as spent)
   (or getting a "you already spent those coins" error from the server)

The lightweight client receives money by:
  Either polling the server every once in a while, asking "Any payments to these BC addresses that I have in my wallet?"
   ... or asking the server to tell it whenever it sees a transaction to a list of BC addresses (or maybe when it sees
    a relevant transaction with N confirmations)
  When transactions occur, the lightweight client updates its wallet (adds the coins).

You don't have to trust the server; it never has your private keys.

Well, you do have to trust that the server doesn't lie about whether your transactions are valid or not, but why would the server lie about that?

[knightmb]

I think then now would be a good point to work out a separate "server" and "client" for the next release. Those want to run servers that sit on fast fiber connections and run 24/7 (like myself) can do this for the good of the service. Those that just want to send money from point A to point B can use the client which either function as a full blown node server (like it does now) or give the option to connect to "trusted" servers in the network and just have the server do the hard work and report back to the client when coins are transfered around properly.

Since the network can be participated by both dedicated server farms and Joe Blow with his laptop, you still maintain the open network and crypto security that you need, but also offer another solution for those less technical and wanting an easy click and send interface that they have now, website payments, etc.

I mean, I'm seeing stories about Bit Coin all over the web now, so the popularity is coming quickly and people are going to poke around the source code and program, then come here to make suggestions/complaints/worship/etc.

Might as well get a head start. 

It's win/win because people can run thin clients, people can run servers, and those that don't trust either can run their own client/server/p2p application themself, but it all benefits BitCoin one way or another. 

[Gavin Andresen]

Making it easier for merchants to accept bitcoins, and users to pay using them, aught to be priority number 1.

There's a great talk by the CTO of Facebook available on Youtube, and I think he gave the right advice on scaling:
Don't worry much about it until just before it becomes a problem.  Don't overengineer, because you're likely to waste time doing something that turns out to be irrelevant.

I think Satoshi has done an amazingly fantastic job; over the last two days of Bitcoin being "slashdotted" I haven't heard of ANY problems with Bitcoin transactions getting lost, or of the network crashing due to the load, or any problem at all with the core functionality.

Yes, it's annoying to have to wait for the block chain to download (especially with the Microsoft Security Essentials weirdness), and yes it would be nice if all the pieces of Bitcoin functionality were already nicely separated and ready to be rearranged and extended in all the ways we all want to rearrange and extend it.  But I've been poking at the Bitcoin code for over a month now, and the more I learn the more impressed I become at the thought that's gone into it.

This quote seems appropriate:
"We reject: kings, presidents and voting.
We believe in: rough consensus and running code."  -- David D. Clark

[spaceshaker]

Making it easier for merchants to accept bitcoins, and users to pay using them, aught to be priority number 1.

Don't worry much about it until just before it becomes a problem.  Don't overengineer, because you're likely to waste time doing something that turns out to be irrelevant.

I agree with both points but they also seem to be somewhat mutually contradictory to me (this may be because of ignorance more than anything). Perhaps you can illuminate what steps you think need to be taken to "make it easier for merchants to accept bitcoins and users to pay using them". In my view, having a lightweight client, build on fairly open standards would facilitate making transactions easier. Ideally this lightweight client would/could be implemented in a number of languages/platforms.

Don't worry much about it until just before it becomes a problem.  Don't overengineer, because you're likely to waste time doing something that turns out to be irrelevant.

This is certainly good advice: "beware of premature optimization". I would like to point out though that optimization and system architecture are different beasts. Planning for scalability is a system architecture task. The whole "Don't overengineer" paradigm is more of an implementation & design principle and suggests that we should be agile in that we only implement what we need today. An agile perspective does not preclude thinking about and planning out the system architecture, at least testing assumptions mentally and ensuring that "it could scale" when the time comes for that ability to be added. I have been involved in a number of projects that failed not because the underlying technology was bad, but because scalability was never considered or tested up front. Everything worked perfectly until the system was put "to the (scalability) test". A lack of planning and foresight up-front was usually the cause.

This quote seems appropriate:
"We reject: kings, presidents and voting.
We believe in: rough consensus and running code."  -- David D. Clark

I like this quote. There is certainly something to be said for having something that works. I tend to struggle with being a bit of a purist; I have to work at pragmatism. Having said this, you can say: "we have great software. It has very few bugs and it hardly ever crashes" and that all sounds well and good until the day it becomes untrue. Scalability issues can hit you quickly and silently, especially in this day-in-age.

My preference would be, for a project like this, would be to continue to have "working software" but also establish a well defined road-map of where we hope the software is going. This road map needs to discuss scalability and how we envision a system with a million transactions a day would work and looks, not just today but 20 years from now. A currency needs to be resilient. For people to assign value to Bitcoin they need to perceive that it has value and that it is safe. This means they need to be able to trust that the infrastructure & design is (and will be) perpetually reliable. Unfortunately I don't Bitcoin, if it is to be successful, as the luxury of being just another hacked together open source project (I'm not implying that it is hacky...I'm implying that the prevailing mindset and practice in open source tends to be a "hack it together" mind set).

[TopSoil]

There's a great talk by the CTO of Facebook available on Youtube, and I think he gave the right advice on scaling:
Don't worry much about it until just before it becomes a problem.  Don't overengineer, because you're likely to waste time doing something that turns out to be irrelevant.

He is also talking about a website. The scalability issues with websites are well known and solved and can be easily tacked on as you get bigger. So yeah you don't need to worry about them when you start. bitcoin is a whole new beast.

Having a collection of servers that act as transaction clearing houses for people would probably work but it seems to break the whole p2p idea of the system. And I think isn't necessary if you design the system right in the first place.

It seems like the best thing would be to lay the system on top of a DHT like kademlia. Maybe you can do this later I'm not sure. Again the documentation is lacking so it is hard to tell if you can easily break the chain apart in order to stick it in a DHT later. I was hoping the devs would stop by and answer maybe they are too busy improving the docs...

[spaceshaker]

Having a collection of servers that act as transaction clearing houses for people would probably work but it seems to break the whole p2p idea of the system. And I think isn't necessary if you design the system right in the first place.

It depends on how it was implemented. I don't think having transaction clearing houses necessarily excludes p2p. It depends on how the network was implemented. Assuming things remained open, the transaction clearing houses could still be p2p. Anybody could join in and work to clear transactions. It's just that it wouldn't be required in order to send/receive transactions. It would actually be extremely beneficial if there were literally thousands of transaction clearing houses or if some regular users did participate in the full system to prevent the established clearing houses from conspiring together to cheat the system. That is one of the beauties of bitcoin IMO.

It seems like the best thing would be to lay the system on top of a DHT like kademlia. Maybe you can do this later I'm not sure. Again the documentation is lacking so it is hard to tell if you can easily break the chain apart in order to stick it in a DHT later. I was hoping the devs would stop by and answer maybe they are too busy improving the docs...

Although DHT is an interesting concept I don't know if it really solves the problem. Smartphones are a good use case:

• You don't want to be acting as a peer on a p2p network with a smartphone. It hurts battery life and it could be costly in cellular data charges.
• Smartphones don't have the processing power or storage capacity to either process blocks or store even parts of the block chain.

The future is not desktop PCs but is "thin" machines like smartphones, tablets and netbooks. These lightweight devices will vastly out-number PCs in the years to come. It isn't tenable to participate in p2p from a thin device, at least in the forseeable future. I think Gavin's ideas of a lightweight bitcoin client seem to address this issue.

[D҉ataWraith]

Having a collection of servers that act as transaction clearing houses for people would probably work but it seems to break the whole p2p idea of the system.

Why? E-Mail and Jabber work the same way. Everyone can run their own server, and many do, but most users prefer to use someone else's server(s).

Edit: Even more closely related: Kazaa and Gnutella work the same way, too. You connect to a supernode that then searches for files on your behalf.

[TopSoil]

Why? E-Mail and Jabber work the same way. Everyone can run their own server, and many do, but most users prefer to use someone else's server(s).

Sure it can work that way but is that the ideal? Doesn't that make the network less robust and more vulnerable to attacks and manipulation? What happens if some attackers start running a cluster of supernodes? The main point is why rely on this more vulnerable architecture when you don't have to? It isn't easier to implement.

Edit: Even more closely related: Kazaa and Gnutella work the same way, too. You connect to a supernode that then searches for files on your behalf.

Yeah and I would argue neither of those systems are very well designed. Limewire replaced the gnutella routing with Kademlia actually.

spaceshaker: yes I agree there will have to be nodes that act as proxies for mobile devices.

[Jman]

Having to download the entire chain also causes issues with new user acceptance.  I installed bitcoin on an older box I have and it took about 4 hours to download the entire chain.  As time goes on this is only going to get worse.  While this is a minor annoyance for most people here with fast connections and servers, if bitcoin were to actually get wider acceptance outside the tech circle it would be a deal breaker.

Say your neighbor wanted to buy something with bitcoins.  You tell him to install the program, then go to the exchange and buy some coins.  Sounds great, right?  How do you think he will feel about the transaction when he has to wait 8 hours after installing the program to get his bitcoins because he has to download a chain with 200,000 blocks?  Do you think he would recommend it to his friends?  To gain more widespread acceptance, initial transactions would need to overcome this factor.

[knightmb]

Having to download the entire chain also causes issues with new user acceptance.  I installed bitcoin on an older box I have and it took about 4 hours to download the entire chain.  As time goes on this is only going to get worse.  While this is a minor annoyance for most people here with fast connections and servers, if bitcoin were to actually get wider acceptance outside the tech circle it would be a deal breaker.

Say your neighbor wanted to buy something with bitcoins.  You tell him to install the program, then go to the exchange and buy some coins.  Sounds great, right?  How do you think he will feel about the transaction when he has to wait 8 hours after installing the program to get his bitcoins because he has to download a chain with 200,000 blocks?  Do you think he would recommend it to his friends?  To gain more widespread acceptance, initial transactions would need to overcome this factor.

I agree, I've been looking into packaging "blocks" for download so that you can install the client, unzip a fresh stack of blocks and let the client finish off the rest to get people up and running faster. I'm experimenting with Windows and Linux clients (don't have a Mac that can run the Mac client, sorry folks)

Hopefully that will help with the issue.

Of course, the trust factor comes in (should I download these from this guy instead of letting the network do it for me)

 

[D҉ataWraith]

Why? E-Mail and Jabber work the same way. Everyone can run their own server, and many do, but most users prefer to use someone else's server(s).

Sure it can work that way but is that the ideal? Doesn't that make the network less robust and more vulnerable to attacks and manipulation?

Not really. If you're worried, you can always run your very own server.

What happens if some attackers start running a cluster of supernodes?

Nothing. Unless you happen to use one of those supernodes, which you don't have to, because you can run your own supernode. Or use a trusted one, much like people trust Google with their E-Mail.

The main point is why rely on this more vulnerable architecture when you don't have to? It isn't easier to implement.

It isn't easier to implement? I'd like to see some proof here.

Counterexample: The totally distributed E-Mail system developed at Rice University is a hell of a lot more complex than running a (network of) semi-centralized mail server(s).

spaceshaker: yes I agree there will have to be nodes that act as proxies for mobile devices.

Um. That's exactly what a supernode server would do.

[knightmb]

I've had luck with testing of packaging blocks into a zip file and dumping them on a new Bit Coin to speed up the block download process. I'll leave the link in my signature and if anyone finds it useful or time saving, feel free to donate. 

[spaceshaker]

spaceshaker: yes I agree there will have to be nodes that act as proxies for mobile devices.

Um. That's exactly what a supernode server would do.

Um. Sure. I think I've gone full circle. I think Gavin said it best:

A lightweight client would have a wallet with coins in it (public+private key pairs).

And a secure way of sending messages to, and getting messages from, any of the ultra-fast, always-connected heavyweight nodes.

The lightweight client sends money by:
  creating a transaction (signing coins with the private key)
  sending the signed transaction securely to the ultra-fast server, which puts it on the network.
  receiving confirmation that the transaction was valid and sent, and updating its wallet (marks coins as spent)
   (or getting a "you already spent those coins" error from the server)

The lightweight client receives money by:
  Either polling the server every once in a while, asking "Any payments to these BC addresses that I have in my wallet?"
   ... or asking the server to tell it whenever it sees a transaction to a list of BC addresses (or maybe when it sees
    a relevant transaction with N confirmations)
  When transactions occur, the lightweight client updates its wallet (adds the coins).

You don't have to trust the server; it never has your private keys.

Well, you do have to trust that the server doesn't lie about whether your transactions are valid or not, but why would the server lie about that?

In this scenario, the Bitcoin client could remain largely the same as it is today, although the focus would be that it is used on the "super-nodes" or "transaction servers" or "proxy servers" (these systems would probably serve all three roles) or by anyone wishing to play in that game. If the Bitcoin client was augmented to use DHT then that may be improvement but there is still a need for a "lightweight client" as Gavin described above. It seem's Gavin's "lightweight client" concept obviates my scalability concerns somewhat.

[BitLex]

And I expect most of us will be running lightweight clients that just keep our wallets, sign transactions, and send and receive transactions to the ultra-fast nodes that ARE looking at every transaction.

Is this possible? What would this look like? From a technical perspective what does a "lightweight client" look like for you? My understanding is that the Bitcoin client needs the entire block chain in order to establish trust.

I'm imagining:
....

you don't even have to imagine , actually it's already possible to "remote-control" the node, so just create ur own little lightweight-client, that just sends and gets some info to/from your (highspeed-connected, hdd-packed) homeserver.

some kind of "managed server-client-version" already exists in MyBitcoin, you could run something like that on your own webhost and connect to it from wherever u are on whatever connection-speed.

the "lightweight client" doesnt have to be one, but connect to it and tell it what todo.
we can do that with JSON.

[InterArmaEnimSil]

I second the DHT idea for maintaining a client list - we can't have millions of people relying upon an IRC channel, etc.  As far as the scaling issue goes, the issue is not at all HDD space, its network bandwidth.  Everyone is forgetting, its not bytes_per_transaction*transactions, which is the number everyone is using.  That number, as everyone has said, is fully manageable.  No, the number we're interested in is bytes_per_transaction * transactions * number_of_clients * total_hops_beyond_first_between_all_clients_combined

THIS is the amount of bandwidth which the protocol for BTC consumes as the network scales.  We're not just talking about sending one copy of each transaction to each client - we're talking about multiple clients broadcasting potentially redundant data to one another, and doing it across numerous hops, meaning numerous rebroadcasts.  Much larger number, much more difficult to handle.  However, it is manageable, just not in the current incarnation of network handling in the client.

Perhaps in the "popular" phase, BTC chains could be broken up by region, similar to the purviews of domain name authorities now - and there could be an alternative protocol for transactions across these regional boundaries?  This would help the raw numbers of the problem, and also cut down on latency and related issues.  Not that I think this is an excellent solution - but P2P flooding across all active clients is obviously out barring some massive breakthrough in quantum computing or whatnot.

[satoshi]

The design outlines a lightweight client that does not need the full block chain.  In the design PDF it's called Simplified Payment Verification.  The lightweight client can send and receive transactions, it just can't generate blocks.  It does not need to trust a node to verify payments, it can still verify them itself.

The lightweight client is not implemented yet, but the plan is to implement it when it's needed.  For now, everyone just runs a full network node.

I anticipate there will never be more than 100K nodes, probably less.  It will reach an equilibrium where it's not worth it for more nodes to join in.  The rest will be lightweight clients, which could be millions.

At equilibrium size, many nodes will be server farms with one or two network nodes that feed the rest of the farm over a LAN.

[lachesis]

It's good the spec already addresses this problem. I'd like to see some more in depth access to things like the coin list (so we can choose to spend coins based upon who sent them to us, etc) and the client list. For example, all of my computers at home -connect to my router, which runs (but does not generate) with port 8333 open.

I wish there were some way to prioritize network traffic to my own local nodes over remote nodes.

[throughput]

I anticipate there will never be more than 100K nodes, probably less.  It will reach an equilibrium where it's not worth it for more nodes to join in.  The rest will be lightweight clients, which could be millions.

At equilibrium size, many nodes will be server farms with one or two network nodes that feed the rest of the farm over a LAN.

Could you, or anyone else, speculate about the numbers of transaction per second limit for the current system?

What could be the maximum transaction throughput (in the number of transactions, not in the volume)? What could it depend on?

What is the future limits on that number, in the case the system will grow?

Just consider Bitcoin as a system for micro- or even nano-payments?
Like paying 0.000001 (or less) for posting in a forum? Or paying even less for just reading the forum (with advertising stripped)?
That would generate far more than a million transactions per day.
That would create unprecedented transaction flow.
Will it hurt the other participants, which are not nano-payers?

We could test the "system" speed if we start sending small amounts in a circle between wallets and thus flooding the system.
Or we couldn't?
If we do that, what effect will it cause?

Would you mind against such an experiment? If you veto that, then how could you stop it?