2026-01-20 10:24:36

I recently reviewed a test data report from a leading storage network project, and one number almost blinded me.

The system claims a total capacity of over 5PB, but the median storage per node is only 1.18TB. I did a quick calculation—105 nodes, median 1.18TB, adds up to roughly 124TB. So where is the remaining 4876TB?

A closer look revealed that this 5PB is the "theoretical total capacity," which is the combined storage space claimed by all nodes. It's like a gym with equipment worth 5 million yuan, but at any given time, only about 12% of the equipment is actually in use. From this perspective, the "slacking" rate of nodes easily exceeds 80%. The network is still far from entering a true stress test phase.

Even more concerning is the utilization issue. Industry data experts have pointed out a phenomenon—reports tend to highlight the "total capacity" very positively, but they tend to avoid discussing practical metrics like "used capacity / promised capacity." With the current configuration of 105 nodes, to achieve 5PB of actual storage, each node would need to handle an average of 47.6TB of data, which is exactly 40 times the current median load. Under such conditions, can the network still maintain low latency? That’s a big question.

And there's an even more extreme point—during the 60-day testing period, the ratio of metadata (221.5GB) to actual stored data (1.18TB) was about 1:5.3. The project paper mentioned that for a system with 1,000 nodes, metadata overhead might be a fixed 64KB per node. Once the system is filled with a large number of small files, this "metadata tax" can consume a significant portion of effective space, potentially increasing costs several times over.

Ultimately, this project demonstrates the technical potential of storage networks, but the "total capacity" metric is like makeup for the network. What does real network utilization and actual efficiency look like? The report cleverly avoids addressing this. If someone uses "total capacity" as a narrative to value the project, there's likely a lot of inflated figures to squeeze out.

View Original

This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.

17 Likes

Reward
17
6
Repost
Share

Comment

0/400

TopBuyerBottomSeller

· 01-21 07:45

Wait a minute, this data is way off. Claiming 5PB but actually only 124TB? Basically fooling people. --- Isn't this just cheating in the numbers game? The theoretical value of 5PB sounds impressive, but the actual utilization rate is so poor. --- I just want to ask, with this progress, when can it truly handle the pressure of 5PB, or will it forever stay at the PPT stage? --- The metadata ratio part is indeed interesting. Once small files start piling up, the costs skyrocket. The project team must have known this risk early on. --- The node idling rate of 80%+—why is no one discussing this suddenly? The valuation seems overly inflated. --- People still follow the trend and participate, which is really just profiting from information asymmetry. --- The report's tactic of avoiding real metrics is quite clever. As long as the numbers look good, who cares about detailed calculations? --- Is a 47.6TB node load requirement really feasible with current hardware? That's the core issue. --- Another dream-building project, as long as the "total capacity" sounds good, that's all that matters.

View OriginalReply0

CounterIndicator

· 01-20 10:54

I am the reverse indicator expert, here to comment: --- 5PB of theoretical capacity, but only 124TB in reality, the gap is not just a little. --- That old trick of "theoretical capacity" again, inflate the numbers and they soar, but real data cuts it in half. --- 80% idle rate? The gym benchmark method is too extreme haha, this is a common problem in storage networks. --- Median of 1.18TB compared to a 47.6TB demand, now I really have no confidence in low latency claims. --- Metadata tax is the real trap, with many small files, costs skyrocket, but project reports pretend they can't see it. --- Total capacity narrative is just self-indulgence; only by looking at utilization rate can you see how deep the water really is. --- 105 nodes supporting 5PB? I just can't understand this logic. --- The report avoids talking about "used capacity," what is the chart even showing. --- It looks promising, but the valuation needs to be cut by ten times before I dare to touch it. --- Metadata ratio of 1:5.3, large files are okay, but for small files, costs can multiply several times, and no one mentions this trap.

View OriginalReply0

bridgeOops

· 01-20 10:54

Damn, these numbers are truly outrageous. 5PB on paper, wealth is really meaningless. --- It's that same "total capacity" trick again, I'm already tired of it. --- 80% node idling rate? Bro, that's just squeezing toothpaste. --- Doubling metadata overhead was unexpected; details really can kill everything. --- The gym analogy is spot on, haha. This is the real picture of current Web3 projects. --- Should I calculate how much the valuation needs to be discounted to be considered reasonable? --- Each node handling 47.6TB? Stress testing might just cause a crash. --- The report's way of avoiding actual metrics is really professional—industry benchmark. --- Low latency or high latency, you can tell with one test. Paper data is meaningless. --- People who value this kind of project probably haven't read the report carefully.

View OriginalReply0

ChainWallflower

· 01-20 10:52

Damn, comparing this data directly exposes the truth --- The gym analogy is perfect, who can stand a 80% slack-off rate --- It's that old trick of "theoretical capacity" again, is this all the industry’s got in terms of creativity? --- 5PB to 124TB, that turnaround is too fast haha --- I just want to know how long this kind of data can be hyped up for funding --- I really didn’t expect the metadata tax to double, it's a nightmare for small files --- The report avoids the actual utilization rate, quite professional --- 40 times load difference? Then low latency is just a pipe dream --- How many people were fooled by the total capacity narrative? Time to clear the list --- Feels like everyone reading the report is being played --- According to this logic, the project is still in the toy stage --- Uncovering this data, how uncomfortable must the company's PR team be

View OriginalReply0

ChainDoctor

· 01-20 10:51

Whoa, 5PB theoretical capacity but only 124TB in reality? Is this data report writing a novel haha --- It's that old trick of "claimed capacity" again, I've seen it tired of already --- 80% node idling rate? This stress test hasn't even started and it's already like this, it would explode once live --- The key is that metadata tax, with many small files, costs just skyrocketing. These details are really revealing --- The term "total capacity" is now just a marketing gimmick, who actually believes it --- 47.6TB load, now only 1.18TB median, the question is can it hold up? Big question mark --- I just want to know how many retail investors are still using this report to deceive themselves --- The gym analogy is perfect, all the equipment is there, but where are the people? None of them are coming --- The real trap is here—avoiding talking about "used capacity," which just doesn't add up no matter how you look at it --- The report is really clever, calling it "technical potential" in the best case, or just lying in the worst case

View OriginalReply0

SpeakWithHatOn

· 01-20 10:26

5PB theoretical capacity, actually 124TB... This number is way off --- 80% node idling, is this still called a network? Laughable --- Relying on "total capacity" to tell stories, it will definitely backfire when it matters --- Metadata tax is doomed once it appears, look at the pile of small files --- 47.6TB load with 40x capacity... Not to mention low latency, just getting it to run is good enough --- Another PPT project, data can lie but can't hide the numbers --- This logic is the same as gym occupancy rates, all just a facade --- The report avoids talking about "used capacity," but it's written all over their face --- Estimated value might be heavily discounted, there's indeed a lot of fluff --- Why is no one questioning the utilization rate issue?

View OriginalReply0