Just out of curiousity, what percentage of traffic on the Tardigrade network comes from customer traffic as opposed to test traffic?
As posted on Aug, 2020. Still waiting for the update though.
I’m curious - what useful data could you take from this information?
As a SNO you will be paid for usage anyway, so from the SNO point of view it doesn’t matter.
From the customer’s point of view it doesn’t matter too - the capacity will expand as you show more demand. Our SNO Community is very pro-active in that matter.
From the token holder’s point of view it doesn’t matter too - the price of the token is usually independent on product evolve unfortunately, it depends on speculations.
It’s just interesting to see how much customer data we have as an indication of “how well” storjlabs is doing. But looking at my ingress patterns I can see that customer data is almost irrelevant compared to storjlabs test data (at least ingress, can’t analyze egress in the same way), which is sad… But with the new team members hopefully there’ll be some more progress soon.
Apart from that you’re right, as an SNO I couldn’t care less where the traffic comes from because I get paid either way.
I’m surprised why you hide this information and provide reasons why SONs should not interested in this information.
It’s not accounted on regular basis that’s simple as it. I’m not hiding anything, it’s just not interested to me.
You can easily calculate on your own data - take a data stored by not customers’ facing satellites and make a proportion to data stored on customers’ facing satellites.
Customers’ facing satellites are listed on https://status.tardigrade.io/
The test data uploaded by customers cannot be uncovered anyhow - data is encrypted.
The goal of my question is to figure out, what the information Community need to have and why is it important. In such a way we together can make a feature request and it can be considered by the team.
On my node:
asia-east-1 1.14 TB us-central-1 1.2 TB europe-west-1 1.49 TB saltlake 5.77 TB us2 7.5 MB europe-north-1 6.59 TB
So, 3.83TB from custoemrs and 12.36TB of test data, with test data making up ~76% of the data in my node.
The problem is, I don’t believe it is this simple. Looking at recent ingress patterns:
Every color in this picture is a satellite (ok mainly 3 at the moment). This is all test traffic (saltlake) even on tardigrade satellites (europe-west and asia). It would be very weird if all customers across both satellites and a test satellite push ingress all at the same time within the same short pulses…
I think there is a lot more test data from storjlabs across all satellites.
We usually using not customers’ facing satellites for tests. But I have no details anyway, so my suggestion is to compare usage from customers’ facing satellites to usage from non-customers’ facing satellites until we could make a feature request.
Generally it’s only interesting to see how well Storjlabs is doing as only customer data generates income/profit for storjlabs. Testdata is partially a means to keep the SNO’s happy, but it cannot go on forever. At some point it’s do or die for the project. As an SNO’s it would be good to now early which one it will be.
Yes, I agree with @tylkomat. It’s also interesting to see how much of the income comes from users vs Storj Labs, even if knowing this proportion doesn’t really affect us SNOs too much. It also helps to gauge the longer-term sustainability.
The percentage of ~one quarter suggested by @Pentium100 is rather encouraging; hopefully, more and more customers will use Tardigrade
I think you do care - as the long term viability of the business is dependent on those customers coming.
It’s simply interesting for storagenode-operators cause its a hint if tardigrade/storj has a future and could be profitable or if its a big bubble and there is no real adoption
Вам это не интересно, поскольку у вас есть взгляд и понимание ситуации “изнутри”. На ответ “вам это не нужно вам все равно платят за тестовые данные” бедным СНО ничего не остаётся кроме как гадать, завтра рухнет сторж или послезавтра? Надеюсь так будет понятнее…
My question is different - what the goal of such information, how useful it could be? Is the useful so great that it worth to distract developers to implement it or is it only “it’s nice to have for curiosity but useless practically”?
We have limited resources and I do not want to distract developers if the implementation will not gain any profit.
Proportion of the customers’ traffic vs test traffic have no sense for me because volume of test data could be reduced, if operators would not increase storage capacity when it is required. This is some kind of insurance of real available space and also we need those tests anyway - to test new features, capture bugs and see behavior of the network, so I cannot guarantee that it could be reduced to zero.
But if we collect opinions why this information can be useful - we can make a feature request and the team may consider to implement it.
I’m personally doubt that this proportion could answer on your question “is Storj could crash? if so when?” (but you asked a little bit differently - “when it crash”, I personally cannot agree with you on that, there is no so much decentralized and end-to-end encrypted alternatives at the moment, if you mean the network). It would be nice to know, what do you mean? The token price, the company or the network?
I personally do not see any relation between proportion in question with any of that.
If you mean when Storj could run out of money - then you can calculate that from the token flow reports:
The price of the token does not related to this proportion too, or even to the evolve of the product, it’s related to speculations.
The volume of the test data cannot say is the network has bugs or is it working normally, so it’s not an answer on question about the quality of the network and could it crash or not. I think that fact that we did not lost any single file since the alpha launch can prove the stability of the network.
So what remained?
Мой вопрос состоял в другом - каково использование этой информации, насколько она полезна - настолько, чтобы отвлекать разработчиков от развития продукта или из разряда “неплохо бы, потому что просто интересно, но никакой реальной пользы нет”.
У нас ограниченные ресурсы и не хотелось бы отвлекать разработчиков, если реальной цели нет. Соотношение тестовых и реальных данных не имеет значения, потому что при увеличении клиентских данных количество тестов будет сокращаться, если операторы не предоставят дополнительное место (правда не гарантирую, что пропадёт совсем - нам всё равно надо тестировать наши продукты, чтобы поддерживать качество и исправлять баги).
Если удастся собрать мнения о полезности этой информации, то мы с вами сможем создать запрос на новую функцию и команда может решить взять её в разработку.
Ваш вопрос поставлен довольно пессимистично - и соотношение не сможет дать вам ответ.
Тем более не совсем понятно, про что вы спрашивали - про цену на токен, про компанию, про сеть?
По моему мнению пропорция не ответит ни на один из них.
Если вас интересует, когда у компании закончатся деньги - то это можно посчтитать по отчётам о движении токенов https://storj.io/blog/categories/tokens
Цена на токен не зависит от продукта, к сожалению, а только от спекуляций.
Количество тестовых данных не может показать, насколько много багов или наоборот всё работает отлично, так что на вопрос про сеть тоже не ответит. И тот факт что мы не потеряли ни единого файла с альфа версии говорит о стабильности сети.
Тогда что остаётся?
@TheoLeeCJ Let’s create a feature request in the Storage Node feature requests - voting - Storj Community Forum (official) and collect the information, why this report could be useful.
i doubt test data will ever be removed completely … like alexey says its required to test the network.
i suppose in theory one could used customer data by some sort of checksum method over the encryption to test the network… but it will never give the same options as being able to have live practical test data on the network which one can experiment and troubleshoot with.
like say one has an object or whatever it’s called that is degraded, how degraded is the data actually, stuff like that one can test with live test data… one can break stuff on purpose just to see what actually happens so one can solve problems before customers run into them.
because in the end like with everything else, it’s about how many different use cases was applied to something and how many times it was broken and rebuilt before something becomes a viable and strong product regardless of what it is…
maybe except for like working with glass or something… but still they would still break it many times in many ways try to make it fail before deciding on a final solution to be mass produced.
different but the same.
and if test data is limit to some satellites and it is that easy to check, then the point in keeping track is a mute point, anyone can from a part of the whole fairly accurately calculate the ratio involved, and then scale it to the amount of data on the network.
and why don’t they like to publish it… well this is a business, doesn’t get easier if new business partners do even the slightest bit of checking and can obviously see they are paying for most of the network, and say… oh well we are a big company we need bigger discounts, we can see you need our traffic to break even…
so even tho this is a point that continually is a question to most SNO’s i don’t really find it so strange they aren’t that interested in having it on their home page or whatever.
also not even a year after launch yet… 25% customer storage is pretty good i think
I can think of some other reasons for patterns to arise. Hourly backups, business hours, maintenance hours. These things are likely to align across unrelated businesses. You’re graph didn’t show the time labels so not sure what kind of period we were looking at. But they do suggest some periodicity.
In general I’d be interested in having some insight into this as well. This would mostly be to judge what organic growth of the platform is in order to make some basic predictions of future growth prospects of our nodes.
I think it’s pretty fair to assume that there will be a limit to test data and at some point the customer data is going to be the vast majority of data on the network.
That said, I’m not entirely convinced the numbers today tell us much about what gains we could expect in the future yet. It’s probably too early days to derive trends from it. Furthermore I personally probably wouldn’t spend developer time on it right now. It would be nice though if we could get some stats in the quarterly town hall with some graphs of how this has changed over time. Both for totals and split between customer data and Storj Labs test data. (I would include test data by customers in the customer data category as it’s paid for by them anyway and generally non of our business whether they are doing tests or having production traffic.)
I don’t think people asking for this are doing so for token speculation and if they are @alexey rightfully pointed out that that’s a fools errand. For me it’s 100% about interest in organic growth of the platform. And I’m pretty sure that’s the case for most others who are interested as well.
Time label was 6 hours. Problem with your theory is that when test traffic stopped, ingress on all other satellites was pretty much non-existent too. So there is a connection.
E.g. look at these past 30 days of ingress. There is always either ingress on multiple satellites (test-satellites and customer satellites), or there is almost no ingress across all satellites. Interesting too: Almost all ingress is from us-central, asia and europe-west, none of those are purely test-satellites. Meanwhile our test-satellites saltlake has no ingress and europe-north only pushes out a lot of repair ingress.
(Note that his graph doesn’t include repair traffic and only shows the real usage netIO)
If StorjLabs have to send large amounts of data to the network’s nodes to make sure space is actually available, that explains why they pay stored data so little
Leaving aside the expansion factor for redundancy… I get the feeling they’re probably not making a lot of profit from data storage. Egress is probably the only profitable operation, for them and for us.
But yes, I’d be curious to know a bit more about how things are progressing too, just because I want to know how well StorjLabs is growing