New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
If he has the skills/capital/stomach to do that...we just don't know.
We use a mixture of owned, leased, and DC provided IPs...depends on the facility and the current need at the time. You can't tell from looking us up...we don't run our own network, we sit on top of our Data Center providers' networks we colocate at. We still buy IP blocks in some cases, acquire some in acquisitions, and rent blocks on longer term deals as needed. Some locations we have unused /24 blocks sitting ready to go. We still have multiple IP leasing sources at affordable prices as well.
.
Hello,
This definitely looks like something we could look into with you we have an abundance of IPv4 available within our MIA location.
Any questions please shoot me a ticket, but we think with your current information we can achieve the sub €1/mo target and have many variants of hardware to assist in the deployment
Wouldn't that be a nice overnight expense doubling... :-)
.
.
I'm not sure if you remember it, but last time it happened multiple providers closed their OVH locations because they couldn't budget the sudden spike.
Francisco
Then why do you still need 1 ipv4 per container? Depending on what you are doing you could just use ipv6 or use nat. Couldn't you reuse one ipv4 for say, twitter and facebook? Or whatever you are trying to do.
It has to be 1 IPv4 per container. Not every application I use supports IPv6 and configuring each of them individually depending on whether they support IPv6 or IPv4 is just too much work.
They also have to work on specific ports which you can’t change.
I don’t know what do you mean by NAT.
Person is running a SaaS with $1000s monthly spend on IPs alone and doesn't care about server location? His customers aren't going to notice 100s of msec ping differences? Tell me another one.
Hints: NAT, ipv6, SNI, port numbers.
I’m not spending $1000s on IPs. Where did you get this information?
This is a distributed network and server location doesn’t matter, so does the ping.
Unbelevebel
anbelevebel
I'm still completely clueless on why each container needs a dedicated IP and why some sort of private IP, forward, NAT (whatever) solution won't work with a HAProxy front end. It appears that each container needs less than 512MB or RAM and of course these are linux/*nix apps (since they run in containers) - so I'm not very clear on why there's no clear port mapping solution esp. since it is very likely that the client IP itself can be easily keyed off for a unique mapping.
But quoting:
nuff said.
@nullnothere He plan to run Masternodes. A masternode requires a public IP at a specific port so that other nodes can connect to it.
The idea of Masternode is to maintain the stability of the network, I am pretty sure making all of them in a single instance defeat the purpose.
Joined April 30 (= yesterday)
This is the answer to all your questions about why do I need so many IPs. Thanks.
and
But I still don't get how you're going to run a master node with <512MB of RAM - presumably you also need to host the full blockchain on it - so even the disk size is going to be big?
I'm not a miner - so I don't have more than superficial knowledge on these things - so pardon any stupidity on my part.
Either way, I assume that not all of this is purely for masternode stuff - so maybe the NATing approach will work for the non-masternode stuff which I assume is still nothing to sniff at volume wise?
It is possible to run on <512MB of RAM depends on the chain that he is currently running on. However, that's on the low end side which might not able to handle with the large amount of transactions.
The most important thing imo about Masternode is stability, I would recommend at least 1GB of RAM to be safe. The disk size is another limiting factor which the OP should take into consideration too. He probably host the new cryptos so the full blockchain is still small (< 5GB).
@nullnothere it depends on the chain. Some (and most) of them takes less than 1GB of DISK space and uses less than 100MB of RAM.
There are some large ones which takes about 5GB of DISK (and this is the largest one I have) and uses around 500MB of RAM.
@FAT32 from my message above and also you've stated that it depends on the chain.
Which means some will need barely any resources and some will need at least decent resources and therefore, building a shared server system will do the job. They will balance themselves.
@FAT32 @nullnothere just to give you guys an idea; I'm currently running about 120 different applications on this Hetzner server and those 120 different chains are taking 126GB of DISK space in total (and that server has 512GB of NVME), using 43GB of RAM and CPU load is 2.30 on average.
As you can see, they are barely using any resources except the RAM. I may still run maybe about 20 or 30 more apps on that server so that would make 150 different chains in total and they would still not able to fulfill the resources that server has, not even close.
That server is €34 per month and running 150 apps on it means 34 / 150 = ~€0.23 cost per container. Except, they only lack unique IP addresses.
Or does it? 😈
@anbelevebel Don't call it a "chain" because most of the cryptos have its own pre-defined port number. Therefore, you can reuse the same IP but different ports for different "chain".
You need some buffer, you can't run production on nearly 100% of all the allocated resources. 100MB is too little for masternodes imo unless you are using the swap memory a lot.
Anyway, if you are offering others this service, please try to provide good uptime.
Thank you for your advice. Do you have an answer for my question in the topic itself? Basically a way to cut down the costs?
If your only concern is about the costs, your best bet will be OVH. Get one of their instance with large amount of RAM then you are good to go.
this sums it up pretty well.
Hi
I can check if one of this fits you: https://www.host-expert.eu/dedicated_servers/best_performance/
When you configure the server you can add up to 100 IPs and on request up to 255.