Next level Bitcoin stress test -- June 29-30 13:00 GMT 2015

[DooMAD]

20 MB blocks means you need 41 Mbps upload. Most people don't have that even available.

Please share your calculations.

I'm apparently missing something.

If we assume that each transaction received is sent to 2 other nodes, then a full 20 MB block over 10 minutes would mean that you would have needed to send approximately 40 MB during that time.

40 MB X 8 bits per byte = 320 Mbit

There are 600 seconds in 10 minutes.

320 Mbit / 600 seconds = 533,333 kbps

That's only a bit more than half of 1 Mbps.

Somehow you ended up with a number that is almost 77 times larger than my calculation.

Does that mean you are assuming that each node sends every transaction it receives to 154 other nodes?  That doesn't seem like an accurate estimate.

What am I missing?

And aside from that, aren't we looking at 8MB now anyway?  The 20MB proposal seems to have been overruled by mining pools.  Granted, there's also talk about doubling it every couple of years, but I expect there will be further discourse about that part.  Even to someone like me who supports larger blocks, doubling could potentially be a little on the excessive side, but then I guess it all depends on adoption levels.

[1715019372]

1715019372

[1715019372]

1715019372

[1715019372]

1715019372

[Luke-Jr]

20 MB blocks means you need 41 Mbps upload. Most people don't have that even available.

Please share your calculations.

20 MB = 160 Mbit.
8 peers minimum.
30 seconds maximum to upload new blocks in a timely manner.
160 Mbit * 8 peers / 30 seconds = 42.67 Mbps.

If we assume that each transaction received is sent to 2 other nodes, then a full 20 MB block over 10 minutes would mean that you would have needed to send approximately 40 MB during that time.

40 MB X 8 bits per byte = 320 Mbit

There are 600 seconds in 10 minutes.

320 Mbit / 600 seconds = 533,333 kbps

That's only a bit more than half of 1 Mbps.

But you can't spend 10 minutes  uploading the block to each peer.
Nor will things propagate reasonably if each peer only uploads to one other peer - consider that SPV/light nodes only download.
The network will just break down entirely...

[loshia]

Here is what I think
https://bitcointalk.org/index.php?topic=766190.msg11769397#msg11769397
In general no mater how big block tx becomes if Tx fee is low there will be always a chance with little money invested to "spam" the network. A reasonable solution for me is to increase tx fee in a first place then increase max block size if needed
That will be a good step forward to be ready when block reward will be arround zero - a good long term solution which will solve most or all of the current issues 

[jbreher]

jbreher, I hope you will perform a test as well. These are very important, to find weaknesses in BTC companies' implementations, etc

My test was tongue in cheek, motivated by a too-short temper. I've since cooled down.

But my point I think is valid. I think there is a chance this 'test' was motivated by less than altruistic scientific curiosity. I was just pointing out that paying a little more than they were willing to per transaction would likely have resulted in prompt validation.

[jbreher]

...makes me think that you are just posting this to get paid for your signature ad.

Please see my sig.

[kano]

... or you could look at the current network requirements of a well connected node and see that the supposedly relevant guessing by a certain person further above is just random vague guessing coz he thinks he knows something but doesn't.

A bitcoin with more than 50 connections averages (well) under 40k bytes a second ... total in both directions ... with the current 1Mb block limit.
It gets above that when you let people download the blockchain from you ... ... so you don't do that if it matters ...

So if that 1Mb limit was suddenly 20Mb and you made the unlikely assumption that it would jump up 20 times ...
you're still under 7mbit ... ... ... total in both directions ... on average

So you'd add on top of that your own transactions you are generating ... to get a reasonable idea of the requirements.

It's like guessing that empty blocks solves the problem of a slow crappy getblocktemplate function someone wrote in bitcoind.
No, it's slow crappy pool code and a badly managed pool that's the problem, the slow crappy getblocktemplate function is minor compared to that.

Look at real numbers ...

[tspacepilot]

because this is like waiting 1h for email delivery..

No, your "mail" is delivered. Its just not confirmed yet.

uhh ok, so the mail is delivered, but I'm unable to read to content and have to wait unknown time, because lot of guys sending mails.. (btw, yesterday I have to wait 5 hours and 28 minutes for first confirmation)

..if you don't feel this like a problem, believe or not, but it IS problem.

More like you can't prove who sent the email for unknown time. Which is actually true, since email is basically always insecure.

Right, but in this analogy, you can't reply or act on the information in the email until you figure that out (no spending coins that aren't confirmed yours yet).

Except Bitcoin does allow spending unconfirmed coins.

Luke-Jr, you would know better than me, but if I see that there's a transaction on the network which spends coins to an address I control, then I guess you're pointing out that nothing stops me from sending out a transaction signed by that address which sends them elsewhere.  Is that right?

But I know none of the wallets I've used allow this ("this will become spendable soon...").   And I'm surprised that a transaction referencing an output which isn't yet confirmed wouldn't be rejected as invalid.

...makes me think that you are just posting this to get paid for your signature ad.

Please see my sig.

Yes, and the coinomat people are starting to get infamous for this.

[Lucko]

20 MB blocks means you need 41 Mbps upload. Most people don't have that even available.

Please share your calculations.

20 MB = 160 Mbit.
8 peers minimum.
30 seconds maximum to upload new blocks in a timely manner.
160 Mbit * 8 peers / 30 seconds = 42.67 Mbps.

BS!!!!!!!!!!!!!!!

Only if you are miner you need that speed. For normal user you can do that in 5 minutes and you are still more then fine. EDIT: "Also who say you will send out to 8 nodes? If you are slower you might send that to only one." Miners are connected directly and have bater connection and don't relay on other nodes. And I have second cheapest option at home and it is 10/100 with no limits(none has limits just to be sure). Is internet really that bad in US that this can be an argument that 40 Mbps is so hard to get? But anyway 4Mbps is more then enough at 20 MB

[elrippo]

20 MB blocks means you need 41 Mbps upload. Most people don't have that even available.

Please share your calculations.

20 MB = 160 Mbit.
8 peers minimum.
30 seconds maximum to upload new blocks in a timely manner.
160 Mbit * 8 peers / 30 seconds = 42.67 Mbps.

BS!!!!!!!!!!!!!!!

Only if you are miner you need that speed. For normal user you can do that in 5 minutes and you are still more then fine. Miners are connected directly and have bater connection and don't relay on other nodes. And I have second cheapest option at home and it is 10/100 with no limits(none has limits just to be sure). Is internet really that bad in US that this can be an argument that 40 Mbps is so hard to get? But anyway 4Mbps is more then enough at 20 MB

Agreed, my full-node has a time offset of -8 at 4Mbps, so this is not the problem 

[Lucko]

I would also like to say thanks for unannounced soft fork of some pools and nodes. Transactions that worked fine 24 hours ago are now rejected by some nodes and miners. The main problem I have with that is that it looks like at lest same of people that are behind this are against announced hard fork. It looks like thanks to this I will have to use my Credit card or cash first time in about 90 days... I just can't get mind funds of a pool... And I see less than half full blocks. So thank you very much for showing me that BTC can be as slow and arbitrary as banks... If a smaller pool operates don't know you will pull this shit that is just great...

[Lucko]

Bigger blocks would - for one - make these kinds of attacks more expensive. This stress test is only slightly effecting others because OP is burning coins as fees and miners are - as its intended - greedy. If the max blocksize is 2 MB the same attack would have needed 40 BTC instead of 20 BTC[1] to have the same effect.

Why do you assume the price of a transaction – the txfee – will stay the same as the supply of transaction space increases?  This contradicts the law of supply and demand.  The price should go down when the supply increases and normal demand stays the same, and the cost of the attack against 2 MB blocks will be exactly the same as the cost of an attack against 1 MB blocks.

Miners can/will make sure of this. If you send 0 fee transation you can see this. Even if blocks are less then full it will take long time to get conformation. Even low fee takes much longer...

EDIT: And we can make a structure that only 1MB is for low fees. Like 50k that is for hi priority transactions... The rest 19MB or 7MB hi fees... Problem solved...

[eleuthria]

Is internet really that bad in US that this can be an argument that 40 Mbps is so hard to get? But anyway 4Mbps is more then enough at 20 MB

Yes.  Most DSL providers in the US only give 768kbps - 1.5 mbps upstream.  It is very rare for DSL to provide higher than that in the US.  If you're in an area with a cable alternative, you can get maybe 3-5 mbps upstream.  Upstream in the US is EXTREMELY low compared to downstream across the board, normally a 10:1 ratio of downstream to upstream at best.

Fiber is rolling out in VERY few areas in the US, and due to the country's size, you're unlikely to ever see it outside of larger cities, not counting the "test cities" that are lucky enough to get rollouts in smaller towns.

Wireless internet (4G) isn't an option because our data caps are also absurdly low.  While many US wireless services advertise unlimited data, ALL OF THEM are lying.  They ALL throttle you down to 3G (and even 2G) after about 5GB of data on their "unlimited plans".  Want more?  Well Verizon can sell you a 150GB 4G LTE plan...if you want to pay $750/month.  All the other carriers are similarly absurd in their pricing if you want more than 5-10 GB/month over LTE.  Oh, and that's up+down combined.

[Lucko]

Is internet really that bad in US that this can be an argument that 40 Mbps is so hard to get? But anyway 4Mbps is more then enough at 20 MB

Yes.  Most DSL providers in the US only give 768kbps - 1.5 mbps upstream.  It is very rare for DSL to provide higher than that in the US.  If you're in an area with a cable alternative, you can get maybe 3-5 mbps upstream.  Upstream in the US is EXTREMELY low compared to downstream across the board, normally a 10:1 ratio of downstream to upstream at best.

Fiber is rolling out in VERY few areas in the US, and due to the country's size, you're unlikely to ever see it outside of larger cities, not counting the "test cities" that are lucky enough to get rollouts in smaller towns.

Wireless internet (4G) isn't an option because our data caps are also absurdly low.  While many US wireless services advertise unlimited data, ALL OF THEM are lying.  They ALL throttle you down to 3G (and even 2G) after about 5GB of data on their "unlimited plans".  Want more?  Well Verizon can sell you a 150GB 4G LTE plan...if you want to pay $750/month.  All the other carriers are similarly absurd in their pricing if you want more than 5-10 GB/month over LTE.  Oh, and that's up+down combined.

So EU is light years ahead of you. Go figure... Relay unlimited 4G in my country cost only 20€/month. 100/100 fibre cost only 42€, 10/40 cable is 35€, not sure about DSL but I guess not too expansive. I pay for 10/100 internet 160 channels+ (including premium like HBO), 72 hours look back option and phone line(calls in ISP network for free) less then 55€.

EDIT: You can get cable practically everywhere and fibre in towns as small as 1500 people. EU did invest a lot in internet.

I guess US needs to slow you down for NSA to look at everything...

[sturle]

Only if you are miner you need that speed. For normal user you can do that in 5 minutes and you are still more then fine.

If you are a very dedicated user, you may accept that your internet is slow as heck for 50% of the time.  Most won't, and will opt for lower security and less privacy with an SPV wallet instead.  I don't think spam is important enough that we should accept more centralization or forcing lower security upon a lot of users.  As the malicious attacks has demonstrated, there is plenty of room for all the non-spam transactions in 1MB blocks.  If your goal is more spam and lower security and privacy, the answer is larger blocks.  If your goal is to make transactions confirm faster, the answer is better spam filtering.

There is another option, if you insist on a hard fork: Smaller transactions.  Using different crypto algorithms, the size of each transaction can be reduced to about 1/4th of the current size.  This unfortunately come with the cost of higher CPU consumption.  Nothing is free.  With each transaction taking 1/4th of the current size, we can do with a block size of 250k for years to come, allowing even users with metered internet and mobile connections to run full nodes.  Assuming spam filtering, of course.  If you want to abolish spam filtering entirely, then no block size would be large enough, and no personal user will even dream of doing a full blockchain download.

[HigsonPP]

What's the transaction fee for it? I think you guys need to clarify all these things from the very beginning ,so it doesn't pose a problem for us. Also i found last test rather interesting. This one was okayish,

[Luke-Jr]

So EU is light years ahead of you. Go figure...

Yeah, and the EU is half the size of the USA...

EDIT: You can get cable practically everywhere and fibre in towns as small as 1500 people.

How about outside those towns?
For example, where there's one hour per hectometre.

[Lucko]

So EU is light years ahead of you. Go figure...

Yeah, and the EU is half the size of the USA...

EDIT: You can get cable practically everywhere and fibre in towns as small as 1500 people.

How about outside those towns?
For example, where there's one hour per hectometre.

Not sure about other states but it is like water or electricity. You will get cable or ADSL or WiFi(EDIT: it might even be fibre) provided by state and you need to only get ISP.

[DannyHamilton]

Not sure about other states but it is like water or electricity. You will get cable or ADSL or WiFi(EDIT: it might even be fibre) provided by state and you need to only get ISP.

In many of those rural areas of the U.S., people have their own personal well for water out of the ground, so water may be a bad example, unless you are saying that their is some way to dig into the ground and tap directly into some high speed internet reservoir?

[Lucko]

What's the transaction fee for it? I think you guys need to clarify all these things from the very beginning ,so it doesn't pose a problem for us. Also i found last test rather interesting. This one was okayish,

It is not a fee problem. It is getting rejected since pool(not BTC mining pool but pays in BTC) is sending to more then 200 addresses. It worked yesterday but now it doesn't. Well to got confirmed after more then 24 hours... So I don't like this soft antispam(what the f**k is spam hire anyway???) fork that doesn't work as it should and it was not announced so we would know how to make transactions that will not be spam according to it. Changing rules like that can kill bitcoin not 20MB blocks.

Only if you are miner you need that speed. For normal user you can do that in 5 minutes and you are still more then fine.

If you are a very dedicated user, you may accept that your internet is slow as heck for 50% of the time.  Most won't, and will opt for lower security and less privacy with an SPV wallet instead.  I don't think spam is important enough that we should accept more centralization or forcing lower security upon a lot of users.  As the malicious attacks has demonstrated, there is plenty of room for all the non-spam transactions in 1MB blocks.  If your goal is more spam and lower security and privacy, the answer is larger blocks.  If your goal is to make transactions confirm faster, the answer is better spam filtering.

There is another option, if you insist on a hard fork: Smaller transactions.  Using different crypto algorithms, the size of each transaction can be reduced to about 1/4th of the current size.  This unfortunately come with the cost of higher CPU consumption.  Nothing is free.  With each transaction taking 1/4th of the current size, we can do with a block size of 250k for years to come, allowing even users with metered internet and mobile connections to run full nodes.  Assuming spam filtering, of course.  If you want to abolish spam filtering entirely, then no block size would be large enough, and no personal user will even dream of doing a full blockchain download.

I don't use core for transactions for a long time. Not safe. If there would have connect a Android device with Trezor option I might use my XT node for transactions but since this will probably not happen MyCelium works just fine. It even has integrated BTC to SEPA in it... So why use a node for transactions? You can't use it of a phone. Big problem if you get hacked. Personally I think suggest that you should use node for transaction is really a bad idea unless you are security expert. Most people aren't. And core is far from secure wallet.

And now what is spam? How do you know that? Who give you the right to decide? If you don't like it don't run a node. Are you a bank? Are we going the way banks are going? And how will users know so they will not be blocked. What happens if someone didn't know that rules have changed? Isn't that why we don't like banks?

And as soon as I see on option that will work without increasing block size limit that will happen in less then 1 year I'm OK with 1MB. But there isn't any... even one that are mentioned need bigger blocks in long run so making them bigger now changes noting. Also I see some conflicts of interests in persons that are against it. Like chef scientist of altcoin or working on some alternative payment system that uses BTC. If your payment needs you not to understand something I assure you you will not understand it. And making BTC settlement layer will make altcoins(p2p users) and that system(BTC users) more attractive. Until now I didn't see an argument that is not BS speculation...

We will need bigger blocks and all alternative payment system, sidechains, altcoins and who knows what more but for now we have size... If you have alternative payment system in place when we need to make blocks even bigger grate but there isn't one now.

And just to be sure. There will not be 2 Bitcoins after hard fork. If it happens 25% will change to bigger blocks or will be killed by bitcoins tht are useless or even 51% attacks for sure... So saying there will be 2 and big problems is BS. When first big blocks is mined small chain will die in hours or days... Unless there are more exchanges and wallets on it. Than it might take longer or it might even manage to beat big one since miners might change back to it. But there will never be 2 Bitcoins for a long period time... Just look what happened with LTC and FTC.

Not sure about other states but it is like water or electricity. You will get cable or ADSL or WiFi(EDIT: it might even be fibre) provided by state and you need to only get ISP.

In many of those rural areas of the U.S., people have their own personal well for water out of the ground, so water may be a bad example, unless you are saying that their is some way to dig into the ground and tap directly into some high speed internet reservoir?

Talking about EU. If you have a building permit you will get electricity, water and a chance to use broadband connection. I should use word country not state.

[achow101]

What's the transaction fee for it? I think you guys need to clarify all these things from the very beginning ,so it doesn't pose a problem for us. Also i found last test rather interesting. This one was okayish,

It is not a fee problem. It is getting rejected since pool(not BTC mining pool but pays in BTC) is sending to more then 200 addresses. It worked yesterday but now it doesn't. Well to got confirmed after more then 24 hours... So I don't like this soft antispam(what the f**k is spam hire anyway???) fork that doesn't work as it should and it was not announced so we would know how to make transactions that will not be spam according to it. Changing rules like that can kill bitcoin not 20MB blocks.

What are you talking about? There was no soft-fork for antispam rules. The soft-fork was for BIP66 which has to do with transaction signatures. Also, a fork cannot be unannounced because the fork only happens when miners mine blocks for it, which can only happen with a code change. They won't change their code for no reason, so fork had to be announced.