elective-stereophonic
elective-stereophonic
Nxt 2.0 design singapore
Please login or register.

Login with username, password and session length
Advanced search  

News:

Latest Stable Nxt Client: Nxt 1.12.2

Pages: 1 ... 40 41 [42] 43 44 ... 51

Author Topic: Nxt 2.0 design  (Read 216075 times)

Jean-Luc

  • Core Dev
  • Hero Member
  • *****
  • Karma: +816/-81
  • Offline Offline
  • Posts: 1610
    • View Profile
Re: Nxt 2.0 design
« Reply #820 on: February 22, 2016, 04:43:17 pm »

@devs: How many transactions per second will be possible in 2.0 on all childchains combined? Is it 256 per minute? Will this number be lower, depending on how many fNXT-transactions we have?
I don't know yet. We will have to increase block sizes and reconsider how block sizes and transaction sizes are calculated too. Because just keeping the same limits, if we count a child chain block as a single transaction and allow a forging chain block to contain 255 of those, this would be 255*255 transactions in a single block at maximum, which it too many. If we apply the 255 limit on the total of child chain transactions, this may be too restrictive if we have many child chains. But before we arbitrarily set the number somewhere in between, the process of sharing blocks between peers must be optimized, for example by only propagating the block headers assuming the peers already have the transactions, and sending the actual transactions only when requested. Then switching from one fork to another will not involve that much network traffic, as the peers most likely already have the transactions in the other fork, it is just how they are grouped into blocks that is different. But the rest of the 2.0 design does not depend on such optimizations, this would be something to work on after the basic framework is ready.
Logged
GPG key fingerprint: 263A 9EB0 29CF C77A 3D06  FD13 811D 6940 E1E4 240C
NXT-X4LF-9A4G-WN9Z-2R322

Marc De Mesel

  • Hero Member
  • *****
  • Karma: +228/-83
  • Offline Offline
  • Posts: 643
    • View Profile
Re: Nxt 2.0 design
« Reply #821 on: February 22, 2016, 04:45:11 pm »

As I've already mentioned in the internal developer email discussions when Jean-Luc first proposed this design, I don't believe this is putting Jean-Lucs awesome coding skills to the best use in the short run for NXT to be successful.

...

Practically every bank and insurance firm out there is looking into 'blockchain' technology and the one thing they are missing (and which is NXT's one mayor chance at world dominatoin - imho) is a standard toolkit to start from

....

Great post Dirk, thank you for sharing.

verymuchso

  • Hero Member
  • *****
  • Karma: +118/-2
  • Offline Offline
  • Posts: 549
    • View Profile
    • HEAT Ledger
Re: Nxt 2.0 design
« Reply #822 on: February 22, 2016, 04:57:04 pm »

I'm aware the following is a little bit like cursing in church but i'm gonna say it anyway, kinda now or never moment.   :'(

NXT has failed as a community (if you'd compare it to Bitcoin).

Hey what would you expect?
Nothing to be ashamed of though still the NXT community is a vibrant one but it's community is not NXT's strong point.

The single most powerful aspect to NXT is it's software and it's developers.
Jean-Luc has done an amazing job, Riker, Ron and the likes have all shown to be excellent developers.

All of which could quit NXT and have a job tomorrow at some blockchain startup or a high paying job at a bank as blockchain consultant/architect.
But they are still here ..

Such a concentration of smart and dedicated people all deeply involved with 2.0 blockchain technology is a powerful thing and the dream of every startup ceo.

Yet the circle of brilliant minds at NXT is not expanding, the group has been the same for a while now. While at the same time over at Ethereum you see the one after the other joining either the core team or some user stepping up and creating and displaying amazing projects.

Why is that?

Simple: Developers (and businesses) can do things with Ethereum.

So should we implement our own smart contracts language tomorrow? Forget about it, not gonna work.

1. That ship sailed with Ethereum NXT will never catch up
2. The amount of time to implement that in a good way simply is not available.

So: What are we left with? Where can NXT truly make a difference, how can NXT be a unique tool and platform in the world right now?
How can we offer what Ethereum cant provide?
How can we provide what Hyperledger is up to now only dreaming of?

Answer: Focus on corporate blockchain use. Accept that its unlikely that we'll ever be paying at McDonalds with NXT.
But do believe that in 5 years 50% of all financial institutions have processes running on NXT based software.

This opportunity is now and it will likely not come around again.

Dirk.
Logged
HEAT: DEX | SDK | HOME

EvilDave

  • Hero Member
  • *****
  • Karma: +341/-40
  • Offline Offline
  • Posts: 1789
    • View Profile
    • NXT Foundation
Re: Nxt 2.0 design
« Reply #823 on: February 22, 2016, 06:08:58 pm »

+lots to verymuchso  ;D
Logged
Nulli Dei, nulli Reges, solum NXT
NXT Donations: NXT-BNZB-9V8M-XRPW-3S3WD
We will ride eternal, shiny and chrome!

Cassius

  • Hero Member
  • *****
  • Karma: +207/-18
  • Offline Offline
  • Posts: 2459
  • Rather be a pirate than join the navy
    • View Profile
Re: Nxt 2.0 design
« Reply #824 on: February 22, 2016, 06:34:29 pm »

Thanks verymuchso. Isn't that exactly what Nxt 2.0 is doing?
Logged
I head up content for BitScan, crypto business hub.

verymuchso

  • Hero Member
  • *****
  • Karma: +118/-2
  • Offline Offline
  • Posts: 549
    • View Profile
    • HEAT Ledger
Re: Nxt 2.0 design
« Reply #825 on: February 22, 2016, 07:00:30 pm »

Thanks verymuchso. Isn't that exactly what Nxt 2.0 is doing?

Sorry. Not following you there.
Don't see how focusing on a complex and costly pruning and sidechain oriented makeover/design that links everything to the main NXT chain is exactly the same thing as ceasing the moment to get out on top as the default commercial blockchain toolkit for financial institutions.

Time and energy can only be spent once. Similar to how opportunities only come by once in a while.
Logged
HEAT: DEX | SDK | HOME

Cassius

  • Hero Member
  • *****
  • Karma: +207/-18
  • Offline Offline
  • Posts: 2459
  • Rather be a pirate than join the navy
    • View Profile
Re: Nxt 2.0 design
« Reply #826 on: February 22, 2016, 07:29:34 pm »

Thanks verymuchso. Isn't that exactly what Nxt 2.0 is doing?

Sorry. Not following you there.
Don't see how focusing on a complex and costly pruning and sidechain oriented makeover/design that links everything to the main NXT chain is exactly the same thing as ceasing the moment to get out on top as the default commercial blockchain toolkit for financial institutions.

Time and energy can only be spent once. Similar to how opportunities only come by once in a while.

Ah, I see.
The pruning/sidechain mechanisms are intended so that white label blockchains can optimally be launched on Nxt.
How would you go about it?
Logged
I head up content for BitScan, crypto business hub.

remix

  • Jr. Member
  • **
  • Karma: +18/-2
  • Offline Offline
  • Posts: 90
    • View Profile
Re: Nxt 2.0 design
« Reply #827 on: February 22, 2016, 07:30:56 pm »

1. That ship sailed with Ethereum NXT will never catch up
2. The amount of time to implement that in a good way simply is not available.

Yes, Ethereum seems to dominate its niche. I see three approaches to decentalized consensus with different main features. I wouldn't expect more than one winner in each category.

1. Dezentralization. Proof of Work, mathematically calculable limits on consensus.
Extremely inefficient. (Bitcoin)
2. Programmability/Flexibility. Can't be optimized/efficient at the same time. (Ethereum)
3. Efficiency/Scalability. No Proof of Work. (NXT)

NXT will achieve full scalability only after version 3.0, according to this thread. That's why I would instead like to see one big update in 2.0 that solves scalability. Now the devs seem to be making incremental changes that are small all things considered, but are big enough to make some people very angry.

Answer: Focus on corporate blockchain use.

Maybe scalability is just the thing corporate use needs.
Logged

websioux

  • Sr. Member
  • ****
  • Karma: +69/-1
  • Offline Offline
  • Posts: 343
  • Great changes grow bottom up
    • View Profile
    • Scriba.io the Blockchain Scribe
Re: Nxt 2.0 design
« Reply #828 on: February 22, 2016, 07:47:10 pm »

+ Moore law double things every 2 years.
- Bitcoin size double every 1 year
- Moore law should hit a wall in 10/20 years dixit G.Moore.
Logged
Secret Miner <= communicate with style | NotBot <= timestamp digital docs

verymuchso

  • Hero Member
  • *****
  • Karma: +118/-2
  • Offline Offline
  • Posts: 549
    • View Profile
    • HEAT Ledger
Re: Nxt 2.0 design
« Reply #829 on: February 22, 2016, 08:13:11 pm »

Time and energy can only be spent once. Similar to how opportunities only come by once in a while.
Ah, I see.
The pruning/sidechain mechanisms are intended so that white label blockchains can optimally be launched on Nxt.
How would you go about it?

1. Forget about the on Nxt, not gonna happen.

Not a single serious financial giant with an IT department of 3000+ developers is ever seriously gonna consider building on NXT.
Higher changes of finding a handful of hopefuls among your own labor force.
And even more likely that they'll be pitched by some startup promising a custom solution (perhaps even one based on nxt  ;) )

2. Pruning is a none issue afaik. Something seriously psychotic is going on in this discussion. On the one hand we want to surpass Mastercard and Visa, while at the same time we want the regular guy to run a chain on his desktop (which is why we must so desperately prune).
Whats it gonna be? Mastercard datacenters or regular guy desktops? Cant do both right?

3. As for white label whatever we call it. All i'm saying is blockchains are here to stay, one or way or the other we are all going to use that.
Wait another year and every mayor enterpise database will have blockchain functionality built in, they are not sitting still.
Perhaps one other year and you can run smart contracts on Amazons custom solution.

But what about now? That handful (but likely many many more) of early pioneers looking for ways to get that competitive edge on their fellow banks/insurers/gambling sites whatever are looking for things to built on.
Simply a blockchain in a box, something open source, extendable. With no lock in just a good platform.

That is where NXT can truly shine.

1. We already have the name (first 2nd gen coin)
2. the code quality is outstanding
3. written in enterpise friendly java, not one big pile of cryptic c, c++ or python

In short. NXT is perfectly suited to be that standard toolkit.

Maybe scalability is just the thing corporate use needs.

NXT is perfectly scalable as is. Just move from H2 to MYSQL or Postgress and we'll be good for years to come.
Down side (or up side).. Hard to run on your RasPi.
Logged
HEAT: DEX | SDK | HOME

Riker

  • Core Dev
  • Hero Member
  • *****
  • Karma: +440/-42
  • Offline Offline
  • Posts: 1796
    • View Profile
Re: Nxt 2.0 design
« Reply #830 on: February 22, 2016, 09:26:24 pm »

@Dirk, I tend to sympathize with your "NXT Enterprise Edition" idea. Let's use the term NXTEE for short.
If NXT would have been a traditional startup then by all means, I think there is money to be made here from licensing, services and maintenance.

However, NXTEE does not benefit the existing NXT stake holders directly, it can benefit the devs for sure, perhaps also the NXT foundation and perhaps can be used to fund future development of NXT under some complex compensation structure.

I would also say that if a big institution would put a large sum of money on the table for such an initiative then I for myself would consider it seriously. Didn't happen so far.
NXTEE may as well materialize side by side with NXT 2.0.
However using this community to promote the NXTEE initiative which does not directly benefit this community is problematic.

I tend to disagree with your comments about MySQL being much more scalable than H2 and use it as a solution.
Reason is database performance is just one factor in scaling, and even if turns out to be the major factor (which I don't think so), I doubt if MySQL will improve significantly.

If you refer to my spreadsheet https://docs.google.com/spreadsheets/d/1C2lgUpSLxUJma3KsJV-lTs0B5mANolZqQZiS8StwVtI/edit?usp=sharing you can see that NXT 1.7 with 200 TPM will generate around 43GB of blockchain data per year. This data has to be shared between 100's and 1000's of nodes. We are talking about blockchain download which will take weeks and requires each node to run a full scale database server.
Logged
NXT Core Dev
Account: NXT-HBFW-X8TE-WXPW-DZFAG
Public Key: D8311651 Key fingerprint: 0560 443B 035C EE08 0EC0  D2DD 275E 94A7 D831 1651

Johnnybtcseed

  • Newbie
  • *
  • Karma: +3/-1
  • Offline Offline
  • Posts: 10
    • View Profile
Re: Nxt 2.0 design
« Reply #831 on: February 23, 2016, 12:05:22 am »

Thank you all for everything. It's been a real fun ride. JL and Riker + team you all are one hell of a good development team. The original vision as I understood it as put forth by NXT creator was that NXT was to do the things that Bitcoin wouldn't/couldn't. The original users of crypto did so precisely because it was not in the same system as traditional controlled banks and corporations. They and their silly money creation + control was always the enemy. So I'm not sure how/why you are now leaning towards a system for these very same people that NXT and Bitcoin was supposed to sidestep.

The good news is that now you don't have to worry about us crazy AE investors mucking up your party. An exit plan from being 100% dependent on NXT AE is underway. This should make your transition to your NXT vision way easier.

Declaration of Asset Independence.....
bitcointalk.org/index.php?topic=1372879.msg13966154#msg13966154
Logged

jl777

  • Hero Member
  • *****
  • Karma: +718/-123
  • Offline Offline
  • Posts: 6170
    • View Profile
Re: Nxt 2.0 design
« Reply #832 on: February 23, 2016, 12:16:41 am »

If you refer to my spreadsheet https://docs.google.com/spreadsheets/d/1C2lgUpSLxUJma3KsJV-lTs0B5mANolZqQZiS8StwVtI/edit?usp=sharing you can see that NXT 1.7 with 200 TPM will generate around 43GB of blockchain data per year. This data has to be shared between 100's and 1000's of nodes. We are talking about blockchain download which will take weeks and requires each node to run a full scale database server.
Short of a believable cancellation of fNXT this will be my last post here and I am sad it has to end this way. I am trying to help NXT, but not fNXT. However it seems that the lure of the fNXT bonus is blinding everybody to believe technical NONSENSE and if basic things like arithmetic are not done correctly, then proper economic analysis is definitely out of reach.

I can multiply and I have common sense, maybe that is all I can do, but I can do those things. The entire justification of fNXT is the bloat, the horrible bloat.

First that it will take weeks to download 43GB. Well, well. exaggeration much? I know there is no iguana for NXT that syncs the entire BTC 60GB blockchain in about half an hour, but even the standard bitcoind does the 60GB in a couple days. As a courtesy I have asked for the network wire protocol for NXT so I could see about adding native NXT support to iguana, but you can imagine the sounds of silence I got from that. Not that there would be the believable BC assurances or technical assistance needed. The key thing to realize is that just because NXT is so slow and inefficient in downloading the blockchain, it doesnt have to be. Maybe that is the problem that needs to be solved?

so it is 43GB/year vs my estimate of 1GB/yr. Let us see which is more likely to be true. The size of the NXT blockchain now isnt 43GB, it is much less, more than 1GB, but about 1GB/yr. That is the entire blockchain. 200 tx per minute might well generate 43GB/yr, but historically we average much less, more like 2 to 4. It was higher when the blocktime was ~2 minutes, now it is a steady 1 minute so closer to 2 per minute.

2 is NOT 100

So this explains the difference of 43GB vs 1GB. But with a bit of work, only the actual transfers and trades of AE along with NXT payments can be put into the fNXT chain. Even I dont mind if all the unfilled bids/asks are pruned (after they are filled or closed!)

So my 1GB estimates factors in a 10x growth in AE activity. Meaning the problem is 43x (or 430x) less than it is stated by the official tech representative, repeatedly. Has my proposal of only storing the net balance changes of assets even been responded to other than with a "no I am not able to do it unless it will take another 2 years"?

Well, maybe java is the wrong language to use if it will take so many years?
Maybe a change of this magnitude should be reviewed by unbiased independent analysts? Since clearly I am not smart enough, find someone that is, pay them 10BTC to study the likely effects of fNXT on the NXT ecosystem. Nope, it has already been decided, better spend the 10BTC on something else.

Now I will be gone, so NXT devs, it will be safe for you to announce and "discuss" whatever you want without worry about me pointing out stuff like 1 + 1 = 2, and that 1 is not the same as 2. And make sure to blame me for anything that goes wrong. I am surprised that I wasnt blamed for the 25% price drop since the 2.0 announcement. If BC2 was my fault, then certainly the 25% price drop must be my fault too? Who cares about the truth? Best to just blame the only one who speaks it. Maybe I am just an emotional child ranting and making no sense? Yes, that must be it.

After all the people who brought you the 43GB and weeks to download estimate claim I am incapable of doing any significant software architecture properly, so I will just go back to completing SuperNET which now will also need an atomic cross blockchain standard for asset migration. SuperNET has a lot of NXT so I was trying to help the others. In fact I think the fNXT hardfork will create a tremendous financial windfall for anybody that understands the complex market valuation issues and is able to create tradebots. So both I and SuperNET will do just fine financially. My warnings are regarding the future of NXT.

[for any economics types, i suggest you study the long term effects of instantly removing half the liquid money supply and locking it away into non-liquid form with a punitive tx fee 100x. hint: it wont be pretty, but hey, it will be to NXT not fNXT so who really cares?]

I havent bothered to analyze fNXT beyond the hardfork as that presupposes NXT is already mortally wounded. But it doesnt matter!

pump, pump, pump the fNXT
to da moon we go.
NAS has no blockchain, who cares?
NXT has no assets, who cares?
pump, pump, pump the fNXT
to da moon we go

James

P.S. for asset holders, do not worry: https://bitcointalk.org/index.php?topic=1372879
you can contact me in slack or BTT, or just ask someone here who is on slack.
 I will do my best to protect interests of asset holders.
Logged
There are over 1000 people in SuperNET slack! http://slackinvite.supernet.org/ automatically sends you an invite

I am just a simple C programmer

lopalcar

  • Hero Member
  • *****
  • Karma: +99/-15
  • Offline Offline
  • Posts: 561
    • View Profile
Re: Nxt 2.0 design
« Reply #833 on: February 23, 2016, 01:38:12 am »

 :'(

I still didn't said anything in this topic, I see pros and cons in both solutions (fNXT with assets and fNXT without assets)
Me, personally, as assetholder, hold assets because they give me dividends, or promise me dividends in future, and the price of this assets is determined by this promises, so if people wants to dump all their assets for get more fNXT, then fine, I will buy much more assets which will give me much faster ROI. "I know, this is a very naive viewpoint, because here there are a tone of speculators and the value of almost everything here is unrealistic and 90% smoke".
I ask myself why so much worry about supernet "or other" assets valuation if people sell them for get fNXT, supernet, ISNTANTDEX, PANGEA... all of them promise dividends according to their usage, not to market price, so if current holders dump them, is a good chance for any real investor to buy this assets, right?

The problem for me, with this solution oriented to scalability is that it will not increasse almost nothing the txs/s as nxt was supposed to do with the famous "transparent forging", I think it's more important to center in a fast synchronization of the blockchain instead of blockchain size at this moment.
What will happen in future when a blockchain based coin needs to achieve 1000 or 10000tps? All of this txs needs to be stored in the forging chain "at least the childchainBlockTransaction", so eventually we will see the same bloat problem arise. Why not implement somehow pruning directly in the current chain?
Really, 48GB per year is nothing, I download films of this size many days and have tones of shit on my HDD, you see that HDD space isn't a problem, people will farm maidsafe same people farmed BURST with their HDD space, there are plenty of them! Just optimice the other bottlenecks... Bitcoin seems not to care about the bloat at this time, just run more light nodes and reward full nodes somehow... Really, HDD capacity will keep growing for many years "I read something about a quartz HDD of many TB "we are years behing this being comercialized, but you see, this doiesn't seems a problem in YEARS"

Another thing which will still match with the original supernet idea "work together among all cryptocurrencies" will be distribute snapshots of the blockchain among other blockchains "if james makes blockchains able to talk between them for atomic swaps and such things this will also be possible", I know, it relies in other systems, but this other systems are also decentralized. CFB also said something about store blockchain snapshots in IOTA transactions "I remember something about some bytes in each transaction for store data"

So that is, there are many other ways to do this and in a better way, because this approach won't solve anything if many child chains are created... we will have problems again later... so please, open your mind to other approaches, fNXT idea is not bad but is far to be good enough. I see it as a well made MS where only some members pays fees in nxt granting the child chain the same security levels as nxt chain, but not as an scalabilty solution, maybe you devs are blind of love with your idea, but look at it from the outside :)

Just say thanks for your work and the time you are spending here devs, only hope you become really open to change your original idea and accept the feedback from community. Take advance that anything have been coded yet and lets choose a better solution among all.

"I didn't read almost anything of the ongoing discussion from this topic for many days, so maybe this suggestions where already presented, just to clarify  :-* "
Logged

MadCow

  • Jr. Member
  • **
  • Karma: +17/-0
  • Offline Offline
  • Posts: 87
    • View Profile
Re: Nxt 2.0 design
« Reply #834 on: February 23, 2016, 02:56:08 am »


... this will be my last post here and I am sad it has to end this way.


James, I am sad to see things come to this point also, but the 'divorce' between NXT 2.0 and SuperNET is probably the best solution now.

fNXT and NXT 2.0 as outlined by Jean-Luc are very exciting and have many advantages for NXT, but isn't good for asset holders for the reasons you have outlined. You should be commended for arguing the case for your many asset holders, but as you suggested, the decision to move to fNXT & 2.0 looks  final.

A continued marriage between SuperNET & NXT under these circumstances can't continue, but an amicable separation is *possible*, and obviously in everyone's interests.

As a holder of NXT and SuperNET assets I am in favor of the divorce, sad though that is. I grew up in a divided house with squabbling parents and in retrospect I wish my parents had separated when it became obvious they wanted to move in different directions. If they had there would have been a painful adjustment period, then they would have felt more in control of their own destinies and i believe they would have been happier people, and made better decisions, and thus been better parents. Their fighting and disagreements made everyone around them miserable. If you stay inside NXT community unhappy with the direction of fNXT you and Jean-Luc will do damage, but if you leave as you have said, then in time you two might gain more respect for each others positions. You will be able to influence your platform and not be restricted to another developers vision, so that's a big plus.

Your Atomic cross chain asset passport is a great initiative! Keep up the good fight for SuperNET :)
« Last Edit: February 23, 2016, 02:59:05 am by MadCow »
Logged

Riker

  • Core Dev
  • Hero Member
  • *****
  • Karma: +440/-42
  • Offline Offline
  • Posts: 1796
    • View Profile
Re: Nxt 2.0 design
« Reply #835 on: February 23, 2016, 06:13:46 am »

If 1Gb blochain takes 24 hours to download then 43Gb will take weeks. Can we improve on this? Probably. You can also download from peerexplorer but this is not decentralized. But to solve this you need pruning.
Logged
NXT Core Dev
Account: NXT-HBFW-X8TE-WXPW-DZFAG
Public Key: D8311651 Key fingerprint: 0560 443B 035C EE08 0EC0  D2DD 275E 94A7 D831 1651

cc001

  • Hero Member
  • *****
  • Karma: +68/-4
  • Offline Offline
  • Posts: 829
    • View Profile
Re: Nxt 2.0 design
« Reply #836 on: February 23, 2016, 07:30:38 am »

If 1Gb blochain takes 24 hours to download

hm?? It indeed looks to me that some numbers and assumptions are a bit unrealistic..?

1. after 2 years of Nxt, our blockchain is 1-2 GB, in one year it should be 43GB? taking into account that we have less and less volume and transactions that doesn't look like a very accurate prediction to me.
2. 24 hours to download 1Gb? Theoretically, a 50mbit line downloads 1GB in 3min. even if you get only 10% it takes only 30min. Why 24h?
3. 100tx/min is about current bitcoin load. Bitcoin, with 6 years of network-effect-boosted growth. With current Nxt growth speed, it will take years until we reach such values.

It sounds a bit like the 1000tx/s talks in the early Nxt days...

On what assumptions do you base those numbers?
Logged
cc001 Personal - NXT-8RXS-2SSK-RNF2-HSNL8
NxtReporting.com - The Nxt Asset Exchange Portfolio Manager with Profitability Tracking - Donations are greatly appreciated on NXT-5W4G-GAR6-JHJP-H8ZTW

Jean-Luc

  • Core Dev
  • Hero Member
  • *****
  • Karma: +816/-81
  • Offline Offline
  • Posts: 1610
    • View Profile
Re: Nxt 2.0 design
« Reply #837 on: February 23, 2016, 07:54:02 am »

The bottleneck is not the network but the disk speed, and the CPU. And the download time does not grow linearly with database size but faster, so those calculations assuming 100x bigger db will take 100x longer are underestimating it.

If you start downloading the blockchain from scratch, it tells you how many blocks per second it is downloading. Initially this number is above 200. Then it drops steadily, and when you are at block 500000 or so, the speed is somewhere around 50-60 blocks per second. (This is on a fast machine with SSD, on a slow one with rotational disk it drops all the way to 5-10 bps). The bigger the database gets, the slower it is to process the next incoming blocks and transactions.
Logged
GPG key fingerprint: 263A 9EB0 29CF C77A 3D06  FD13 811D 6940 E1E4 240C
NXT-X4LF-9A4G-WN9Z-2R322

testdruif

  • Full Member
  • ***
  • Karma: +71/-1
  • Offline Offline
  • Posts: 223
    • View Profile
Re: Nxt 2.0 design
« Reply #838 on: February 23, 2016, 08:04:08 am »

One "small" question.

Right now a block is forged via a "lottery system" where the ones with the biggest stack of nxt create the block.

Imagine that I want to have a childchain block included and am only willing to pay 1 fnxt and this price is "too low" according to all forgers (in a world where forgers strong arm people to pay more fnxt).

Could I, with my (lets say) 5 fnxt, forge that block myself thus providing security for my own transactions?
Logged
**Necessity is the mother of invention**
NXT-NNGD-V8TN-3MZR-DWWBE
https://arguseyes.net

Riker

  • Core Dev
  • Hero Member
  • *****
  • Karma: +440/-42
  • Offline Offline
  • Posts: 1796
    • View Profile
Re: Nxt 2.0 design
« Reply #839 on: February 23, 2016, 08:13:40 am »

If 1Gb blochain takes 24 hours to download

hm?? It indeed looks to me that some numbers and assumptions are a bit unrealistic..?

1. after 2 years of Nxt, our blockchain is 1-2 GB, in one year it should be 43GB? taking into account that we have less and less volume and transactions that doesn't look like a very accurate prediction to me.
2. 24 hours to download 1Gb? Theoretically, a 50mbit line downloads 1GB in 3min. even if you get only 10% it takes only 30min. Why 24h?
3. 100tx/min is about current bitcoin load. Bitcoin, with 6 years of network-effect-boosted growth. With current Nxt growth speed, it will take years until we reach such values.

It sounds a bit like the 1000tx/s talks in the early Nxt days...

On what assumptions do you base those numbers?

1. Please take a look at https://docs.google.com/spreadsheets/d/1C2lgUpSLxUJma3KsJV-lTs0B5mANolZqQZiS8StwVtI/edit?usp=sharing
200 TPM is not where we are today, it's where we want to be.
Sure if we continue at the current rate of 2 TPM then all these pruning is not necessary but we want to scale.

2. Downloading the blockchain is not like downloading a file. You need to download blocks in chunks of less than 720 at a time from multiple peers which need to validate each others data. It's more like torrent download with redundancy. We are already using web sockets instead of Http for this process but we can probably optimize if further with a lot of effort.
On slow VPS node this can take 24 hours. On a stronger machine can be done in 2 hours.
You can always skip all this and download everything from a central server but then you compromise on security.
A compromise of slower but more secure solution would be to download a blockchain database with the last 1440 blocks popped off and without derived tables then let the node rebuild the derived tables and download the missing 1440 blocks ((or download the full thing then locally popoff 1440 and rebuild the derived tables). This will take less time than download from start but still building the derived tables on a slow machine currently takes hours and will take days of spinning the disk with 43GB blockchain.
Logged
NXT Core Dev
Account: NXT-HBFW-X8TE-WXPW-DZFAG
Public Key: D8311651 Key fingerprint: 0560 443B 035C EE08 0EC0  D2DD 275E 94A7 D831 1651
Pages: 1 ... 40 41 [42] 43 44 ... 51
 

elective-stereophonic
elective-stereophonic
assembly
assembly