How Ready Is Hive to Handle Bursts of Activity?

in #hive-167922last year

After Splinterlands took off incredibly at the end of the last bull market, we could hear stuff like: "We need 100 Splinterlands for Hive to really shine".

That might be true, and we might shine with 50 or 10 "Splinterlands", but let's not forget what happened during the incredible growth of Splinterlands.

I may be reporting them in the wrong order, but all of them happened in the course of a few months while the number of accounts and activity of Splinterlands grew exponentially.

First of all, Splinterlands ran out of account claim tokens relatively quickly. While they implemented a different solution, I believe other holders of major account claim tokens were creating accounts for Splinterlands for a short period.

Their solution was to remove account claim tokens from the equation completely because this only works when the flux of new accounts created is predictable. It will lead to blockage if a huge number of accounts is created in a short period of time, which is what happened during that period.

The advantage of Splinterlands was that it had a product (the Summoner's Spellbook), which was transformed into a gateway to creating a new account. And the price for that product was enough (10$) to cover the creation fee for a Hive account. Therefore, at that time they burned 3$ worth of HIVE to create every account that purchased the spellbook.

I'm not sure if now that the flow of new users has slowed down, the option to burn HIVE to create new accounts is still used or if they are using account claim tokens again. But if the craziness starts again, they have the option already built in.

What else happened back then? Despite Splinterlands having their own nodes, as far as I know, the blockchain (nodes) simply couldn't process as many transactions as they were pushing in a manner that didn't degrade the player's experience to a great degree.

Two things happened. First, Splinterlands took off-chain all game-related operations and left on-chain only market operations (all operations involving assets).

That pretty much solved the problem.

The other thing that happened was that Blocktrades' team discovered an issue in the core code that would prevent some operations to be handled in parallel, as they should. They fixed that.

hand-4372144_1280.jpg
Source

Why the history lesson?

Because we are about to repeat it.

This time it's not Splinterlands, it's Leofinance, and it's not a huge number of users onboarded yet, but it is a significantly increased activity level.

It is very important to think this through for scaling because it happens at 1000 or so users. Splinterlands went from 20,000 to 500,000 MAU in the course of a few months.

It is also important to take into consideration the differences between Splinterlands and Leofinance (Threads, in particular, for the activity).

One difference is that, unlike Splinterlands, it makes little sense to have multiple accounts on Leofinance (Threads), unless you split them by categories of content. But the impact on the nodes for one very active account from Threads is higher than it was for Splinterlands when it had everything on-chain. Because a comment is heavier on resource credits than a custom JSON.

I just listened to the Leofinance AMA. Looks like because of the increased activity, API nodes are adding limitations to Leofinance (i.e. Threads). Khal was considering rebalancing the load between multiple nodes or having their own node or both.

Looks like some decisions need to be made and whatever they do, it will probably be a stressful period for them.

Let's return to the question in the title.

Is Hive Ready to Handle Bursts of Activity?

I believe the answer is a clear no. But within the crypto-sphere, Hive is one of the most prepared ones, already being a top 3 blockchain by activity for a long time, and with room to scale up.

To be fair, I don't believe any business can be ready for a sudden burst in activity or sales. That's why preparations are needed in anticipation, collaboration, and understanding when things start to break because of such a situation. Plus swift action when it hits you, usually at the worst time for you.

Posted Using LeoFinance Alpha

Sort:  

I agree it could handle more Splinterlands if they kept most operations off chain except for assets. But to really be more valuable another fifty major games or even businesses would be needed but they would likely also test the system. But that's a good thing so it can grow!

That's true, pushing the limits often results in adapting to growth. Otherwise, complacency in a certain status quo would lead to no or little progress.

Nice share. Hope infrastructure can catch up with activity levels so that people will not get disappointed or frustrated.

That's the idea. When something like this happens, you can only play catchup. And UX will always suffer. Nobody on a budget pays for an infrastructure that supports potential huge spikes in activity if that activity isn't there to justify the costs. And when activity spikes, it takes time to catch up. And users might get frustrated.

I can easily say, hive can easily handle this kind of activity in the coming time.

We have to understand Hive is, at a technical level, a network of nodes processing transactions. If transactions are chocking nodes at any given time, they have these options:

  • upgrade their infrastructure (costs more money and few will do it unless there is a clear direct benefit to the node operators that will help them cover the extra costs)
  • limit the type of transactions or the source of transactions accessing the nodes (results in delays and failures)
  • take the node down or make it private if the costs become unsustainable

Also, Hive's base code reacts by increasing the resource credits cost of that type of transaction, regardless of the source. If we are talking custom JSONs, a very popular Splinterlands compared to today should increase the resource credits costs for custom JSONs for everyone using them. Likewise, a very popular Threads would likely somewhat increase the resource credits cost for posting & commenting in general. I've seen this looked at with good eyes, as it would force applications to buy and power up HIVE or rent resource credits for their users.

Twitter is ready for exponential user growth!

No, it's not.

@gadrian is that the real Elon Musk? So interesting.

Posted Using LeoFinance Alpha

I doubt it. But whoever he or she is, seems to have taken an interest in Hive and how it works. Enough to understand some of its problems too.

OK thanks @gadrian 😆 !BBH !CTP

Interesting post, Adrian.

I was not around for the Splinterlands launch but I hope that the powers to be have a plan in place for a sudden surge.

If someone comes over from Twitter and experiences delays they may very well leave and never give Leo Finance a chance.

Just my thought.

!CTP
!ALIVE

@gadrian! You Are Alive so I just staked 0.1 $ALIVE to your account on behalf of @beststart. (5/10)

The tip has been paid for by the We Are Alive Tribe through the earnings on @alive.chat, feel free to swing by our daily chat any time you want.

If someone comes over from Twitter and experiences delays they may very well leave and never give Leo Finance a chance.

That may very well be true. It is a double-edged sword to have short-form content on Hive. You have censorship resistance and account ownership from Hive, plus the potential rewards, but you also can run into scalability issues more frequently. This isn't necessarily a bad thing, as it tests the limits of the blockchain, and eventually pushes them.

@aggroed is building the dacter dapp which would allow for trading account tokens and would make it probably much cheaper than $3 to trade them. This is one hope I have, but not sure where the project is at atm. Otherwise, big accounts should claim all the tokens they can. If we run out, Hive will very fast realize that it actually can't scale with activity as no one will pay the $3 for creating an account

@tipu curate

I hope that that app has a future. I haven't heard him talk about it since the time he announced it.

Yep, account claim tokens and the fact they are not tradable (but even if they would be), is one clear thing that can limit the speed of user growth when we talk about big numbers. I believe it's by design, much like the market cap of HBD doesn't go to the sky except if HIVE goes to the sky too.

That's why I've said:

Fast Feeless Scalable

Has not been achieved because in reality the blochain is throttled and LEO found out the hard way that HPE don't see his endeavor of comment spam disguised as 'Twitter On Chain' as valuable content. People are literally asked to add x amount of threads then hit ATHs outdoing each other spamming the chain.

You and LEO community need to think about that. I made that point long ago to Khal that likely HPE would consider the 'Short Form Content' activity as shitposting. He said he also thought on it but had measures to keep operational health.

Looks like not.

Anyway another point I made was that staying dependent on the development and resources of others was a point of failure for LEO but I was labeled a hater and got downvoted.

I also said LEO wasn't ready for defi and should concentrate on community building. 2 years later DeFi flatlined and community building is exactly what LEO has gone back to.

GREAT MATE
UI LOOKS GREAT.
BUT CONTENT FATE ?
NOW THE CONUNDRUM !

This post has been manually curated by @bhattg from Indiaunited community. Join us on our Discord Server.

Do you know that you can earn a passive income by delegating to @indiaunited. We share more than 100 % of the curation rewards with the delegators in the form of IUC tokens. HP delegators and IUC token holders also get upto 20% additional vote weight.

Here are some handy links for delegations: 100HP, 250HP, 500HP, 1000HP.

image.png

100% of the rewards from this comment goes to the curator for their manual curation efforts. Please encourage the curator @bhattg by upvoting this comment and support the community by voting the posts made by @indiaunited.