I find the server prices by us in Germany very high if storage needs to be included. There are cheaper countries, but Germany has very good connections. (In the cities xD)
well actually there should be a difference so far as i have been able to tell, storagenode incoming and outgoing data atleast mostly for now… is almost like clock work… if you compare two nodes of the same age they will have nearly the same amounts of data on them and upload and downloads will be basically identical within a 5-10% or so margin… which is most likely down times and what not…
the satellites seems to allocate the bandwidth and data… thus a 1gbit node may fill faster… in theory… but when you reach the level of the other nodes of the same age and reputation or whatever… then you will follow along with the pack basically…
atleast so far as i have been able to gather from trying to study the subject without going all science on it…
you can even see it if you are offline for a few hour or so… then the sats will increase your allotment when you get back on… was a bit of a puzzle for a long time for me… because whenever i would test then i might get variation over short periods…because of downtime…
making it very difficult to do short term tests… when the sats get either pissed that you keep rebooting or allot you more data for expended downtime.
ofc i also assume unlimited bandwidth and unlimited storage space… for this…if somebody has like minimum required bandwidth… maybe they will see larger differences… but i kinda doubt it… they will just seen uploads over longer periods where the faster connections are idle.
yes, of course. However, it doesn’t makes a lot of sense: partially canceled messages are not valid:
Is it? Been running on a CIFS share for 2 months now.
It is. The SQLite is not compatible with any network protocols, except iSCSI. Sooner or later you could have a problem with sqlite databases.
The only exception could be if CIFS is shared from the Windows server and the client is Windows too.
You can take a look: https://forum.storj.io/tag/smb
don’t see how that relates to satellites controlling the allocation of data on nodes…
what i was trying to say is if nodes have the same age, no downtime, infinite bandwidth and infinite storage, then they would most likely be 99% the same in data stored and bandwidth used for egress.
atleast for test data which is what we see the most of… but i suspect normal data will be distributed much in the same way… that way it’s easier to keep the network balanced and stable… actually makes perfect sense when you think about it…
i tried using network drives, but didn’t have much luck with it… also it might add latency… ofc it depends on a lot of factors… stuff is rarely impossible, just very difficult xD
i did design my setup to hardwire up to like 1000 hdds into the bus of my server so no need to iscsi, cifs and what not… ofc there will be a bus speed limitation at one point… but i’m betting the internet goes first lol
Thank you for sharing.
Your success rate with the pi3 is about the same with my pi4. For me, I think the upload speed is the bottleneck. I think then I will put the pi3 to my parents’ house. If afterwards it’s $ 2-3 a month, I’m happy. The hardware is already there and will not be used anyway.
Hi sir I just wonder how did you do 24tb on just 1 node?
You didn’t ask me, but RAIDZ1. I just happened to know.
I use SHR2 on Synology myself.
Hi sir, Im also newbie, what do you mean by node gets vetted? I also have 3.2tb node
You didn’t ask me again, but since I’m at it, haha. Nodes need to succeed 100 audits to get vetted on the corresponding satellite. Until they are they only receive 5% of normal traffic. You can use the earnings calculator to see the progress of vetting on each satellite.
i’m using zfs… but in short i combined a number of drives in a raid solution to get some redundancy and other performance benefits for a large array / pool which i will be using for a wide range of things, like vm’s, storagenode/s, home media storage, network drives and pxe server
and to be more exact.
its currently a pool of two raidz1 vdevs, and when my scrub of the pool is complete i will be adding a 3rd raidz1 vdev to amp up my raw iops the pool can handle.
capacity operations bandwidth pool alloc free read write read write --------------------------------------------- ----- ----- ----- ----- ----- ----- rpool 55.0G 84.0G 0 47 0 637K ata-OCZ-AGILITY3_OCZ-B8LCS0WQ7Z7Q89B6-part3 55.0G 84.0G 0 47 0 637K --------------------------------------------- ----- ----- ----- ----- ----- ----- tank 13.6T 11.0T 2.03K 153 803M 3.63M raidz1 8.13T 8.23T 769 41 406M 1.73M ata-HGST_HUS726060ALA640_AR11021EH2JDXB - - 249 13 136M 590K ata-HGST_HUS726060ALA640_AR11021EH21JAB - - 219 14 135M 590K ata-HGST_HUS726060ALA640_AR31021EH1P62C - - 300 13 135M 590K raidz1 5.43T 2.74T 1.28K 52 396M 577K ata-TOSHIBA_DT01ACA300_531RH5DGS - - 586 18 132M 193K ata-TOSHIBA_DT01ACA300_99PGNAYCS - - 578 17 132M 192K ata-TOSHIBA_DT01ACA300_Z252JW8AS - - 147 16 132M 193K logs - - - - - - ata-OCZ-AGILITY3_OCZ-B8LCS0WQ7Z7Q89B6-part5 63.7M 4.44G 0 58 0 1.34M --------------------------------------------- ----- ----- ----- ----- ----- ----- temp512 9.03T 1.88T 0 0 0 0 ata-HGST_HUS726060ALA640_AR31051EJS7UEJ 4.55T 921G 0 0 0 0 ata-HGST_HUS726060ALA640_AR31051EJSAY0J 4.47T 1002G 0 0 0 0 --------------------------------------------- ----- ----- ----- ----- ----- -----
and looks like it will be fast fast fast lol 3300 reads of a total 800MB pr seconds, and writes should be not far removed from that…and ofc + 50% better in a bit… ofc will take a while to rebalance the data, but didn’t have enough drives to actually have them all in the pool when creating it and migrating my storagenode…
but thats just a balance issue, should solve itself even… if i don’t figure out how to make it rebalance.
and because i don’t have enough 6tb drives it will be a mix of 3tb and 6 tb lol
me ghetto rig
1 node - dedicated NAS with 4 drive bays, currently on raid 1 (SHR1 with 2 HDDs).
10 TBs total storage.
However knowing what I know now, I would have just gone with a cheap Rpi and a 10 TB drive since they would earn the same and the rpi would be cheaper as an initial investment.
SNO’s with higher end gear don’t get paid any different to lower end gear.
A post was split to a new topic: iOS/Android/Mac/Win App like dropbox
Hi! Out of curiosity, would you be willing to share the name of some of these projects? I have a very quick internet connection and plenty of storage that I would like to put to good use. If not, no worries. Cheers. Tiago
14 TB on 3 nodes not full on 2 public IPs with the same provider (5$/m option for labbing)
2 x2TB old nodes full running on corei3 mini-pc hackintoshed on applefs as external USB devices
A macmini i7 as cold standby configured to takeover mini-pc (no failure so far) used for other minings
10TB on synology 412+ in Raid5 not full yet but very fast
All recycled materials with no crash. Just had to slow down the ingress throughput to the minipc no accepting more than 3MBs due to low specs
Planning to start a rapi4 one in july to learn docker ARM
Looking forward the syno backup app to recycle my STORJs AND consolidated Https dashboard
If you are ready to spend time on projects not in prod, Check out golem or Akash network which in currently in challenge phase2 for building a decentralized cloud infrastructure with compute and storage.
How do you limit the ingress ?
6 posts were split to a new topic: Is it possible to run 2 docker node on one pc and how is it done
Siacoin is the only other publicly participate-able project, but I price SC/TB stored like Storj pays but bandwidth a lot cheaper, do not get much for contracts as a result, prices for it are rock bottom with lots of people providing storage and bandwidth for much cheaper. My other projects are just some other non-decentralized side gigs.