Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!


Shells Virtual Desktop
BMail.ag - Secure Email Service
Server.net
CPLicense.net
VPS Server
Buy VPN
Vultr
VMs for AI
HostDare
ReliableSite White-Label Dedicated Hosting for Resellers
InterServer VPS
BMail.ag - Secure Email Service
Best VPN
High-Performance Bare Metal Server Solutions
Karvl.com
Server Mania Cloud Hosting
DataWagon Hosting
AlphaVPS Hosting
Evoxt.com
Clouvider
VPS Hosting with NVMe
Residential IPs in the US & 4G Mobile Proxies in EU & US with Unlimited Bandwidth
ReliableSite White-Label Dedicated Hosting for Resellers
Rabisu - Hosting Solutions
Shells Virtual Desktop
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.

★ VirMach ★ RYZEN ★ NVMe ★★ $8.88/YR- 384MB ★★ $21.85/YR- 2.5GB ★ Instant ★ Japan Pre-order ★ & More

12425272930339

Comments

  • RocksterRockster Member
    edited March 2022

    @VirMach said:
    176.119.148.0/22
    78.142.228.0/22
    88.214.20.0/22
    213.232.112.0/22
    147.78.240.0/21
    45.66.128.0/21

    Oh, just geolocated RIPE IPs. Hoped for proper LIR IPs from APNIC region when I saw comment about IPs provided directly from the colocation provider.
    Even after various databases updates some local services which require proper LIR IPs won't ever recognize them as proper Japanese IPs.
    Two years prepayment went straight down the drain. What a shame.

  • VirMachVirMach Member, Patron Provider

    Two years prepayment went straight down the drain. What a shame.

    You can still apply for a refund in most cases.

    Thanked by 1Rockster
  • foitinfoitin Member
    edited March 2022

    @VirMach said:

    @foitin said: In other words, virmach bought 10Gbps dedicated (or shared?) port from xTom and they won't pay extra unless all virmach Tokyo users consumed more than 10 GB/8 * 60s * 60 min * 24 hr * 30 days = 3240000 GB bandwidth?
    More than 3 PB does look like overkill but I'm more concerned about the speed.

    The math is wrong on this here, that's not what I meant either.

    If it's wrong then it's probably beyond me and I might not able to understand how it works.
    Could you bear with us a little more and show us the correct calculation and what outbound speed can we expect from Tokyo NVMe and storage VPS? (of course I'm talking about traffic to areas other than China so that firewalls and state-level throttling won't be an issue.)

    Right now with the number of signups we have, if every single person used all of their monthly bandwidth, we still wouldn't run out. So there's no reason to even begin worrying about any of that.

    Actually most of us aren't worrying about bandwidth, but rather the speed. Your mentioning 40Mbps made us worried for a moment.
    I expect much faster speed to domestic residential network to say the least :)

    I apologize for discussing the amounts that actually get used on the back-end. Even as a provider it's hard for me to believe how little bandwidth actually gets used. There's a reason providers "double bandwidth" or provide unmetered bandwidth, because probably 99.9% of users will not actually even get close to using 1TB of bandwidth per month.

    IMO unmetered bandwidth or over-allocated bandwidth isn't beneficial to users as a whole as it encourages abusive behaviors like torrenting and all other users on the same node are affected.
    When users can only use up to, say 1TB per month, they calculate and save bandwidth on trivial things. This is especially true for storage VPS where torrenting and streaming are made possible.

  • @VirMach said: I know this is difficult to believe as it is for me as well, but when I said 40 Mbps I specifically meant Megabits, as in 5MB (Megabytes) per second. For all the infinite bandwidth providers hand out, only about that much seemingly gets used per node and this is on our beefy 256GB RAM + dual Xeon servers.

    For me, it was so hard to believe how little transfer you use, I originally thought you had made a mistake with the 40 Mbps per node. I guess most people do buy a VirMach VPS to idle it. :)

    @VirMach said:

    @FrankZ said: NOTE: This is my 3,000th post on LET

    Congrats! I wonder how many of them are on VirMach threads.

    I would expect more or less half of them. I am working out how to get the Vanilla search function to tell me, as I would like to know this too.

  • NoCommentNoComment Member
    edited March 2022

    @foitin said: If it's wrong then it's probably beyond me and I might not able to understand how it works.
    Could you bear with us a little more and show us the correct calculation and what outbound speed can we expect from Tokyo NVMe and storage VPS? (of course I'm talking about traffic to areas other than China so that firewalls and state-level throttling won't be an issue.)

    https://www.reddit.com/r/networking/comments/fkjelj/datacenters_burstable_bandwidth/

    The way they are paying for bandwidth is nothing new for colocation. You get a 1 Gbps shared port on the ryzens and 10 Gbps shared port on the storage servers. That's the simple answer.

    Thanked by 1plumberg
  • @NoComment said:

    https://www.reddit.com/r/networking/comments/fkjelj/datacenters_burstable_bandwidth/

    The way they are paying for bandwidth is nothing new for colocation. You get a 1 Gbps shared port on the ryzens and 10 Gbps shared port on the storage servers. That's the simple answer.

    Thanks. This is clear enough.

    Problem is, how could 10Gbps be enough for 800*4/2 users without throttling assuming all NVMe plans are sold out and half users chose Tokyo.
    e.g. 1Gbps (both UL and UL) is the minimum speed for Japan FTTH/FTTB and there're 2Gbps, 5Gbps even 10Gbps residential network. The same can be said for Korea, Hong Kong, Taiwan and Singapore to say the least.
    MMJs' speed is capped by their government but in general 1Gbps is quite common in East Asia.

  • @foitin said: Problem is, how could 10Gbps be enough for 800*4/2 users without throttling assuming all NVMe plans are sold out and half users chose Tokyo.

    They have statistics from years of operations and they determined that average usage was 40 Mbps per server on 256 GB ram e5s. 10 Gbps will probably be enough most of the time. The odds of multiple people rsyncing their entire drive at the same time are low, and even if it happens at least you can share the bandwidth.

    What use case do you have that you require consistent 1 Gbps uplink? In such a situation maybe you should consider dedicated instead.

    @foitin said: e.g. 1Gbps (both UL and UL) is the minimum speed for Japan FTTH/FTTB and there're 2Gbps, 5Gbps even 10Gbps residential network. The same can be said for Korea, Hong Kong, Taiwan and Singapore to say the least.

    And in these situations the bandwidth is not guaranteed. Typically, the contract would mention how much bandwidth is guaranteed. It could be something like 10-20 Mbps guaranteed international bandwidth even if you have a "1 Gbps plan". But you still get decent connectivity because most people are not using their full port speed 24/7. The same logic applies here.

  • @foitin said:

    Problem is, how could 10Gbps be enough for 800*4/2 users without throttling assuming all NVMe plans are sold out and half users chose Tokyo.
    e.g. 1Gbps (both UL and UL) is the minimum speed for Japan FTTH/FTTB and there're 2Gbps, 5Gbps even 10Gbps residential network. The same can be said for Korea, Hong Kong, Taiwan and Singapore to say the least.
    MMJs' speed is capped by their government but in general 1Gbps is quite common in East Asia.

    From the data of my vpn provider, 10G for 2000 people is pretty alright.
    I would say you can get at least 500Mbit most of the time, perhaps 200 in peak hour.

    I don't understand why this concerns you, just look at the price.

    Lower your expectations, you are looking at something that's 5x cheaper than average.

    Thanked by 1kheng86
  • @NoComment said:

    What use case do you have that you require consistent 1 Gbps uplink? In such a situation maybe you should consider dedicated instead.

    Backing up whole disk, not frequently but hope burst speed is decent.

    And in these situations the bandwidth is not guaranteed. Typically, the contract would mention how much bandwidth is guaranteed. It could be something like 10-20 Mbps guaranteed international bandwidth even if you have a "1 Gbps plan". But you still get decent connectivity because most people are not using their full port speed 24/7. The same logic applies here.

    Yeah. I read the contract it's basically best effort but from my experience, I get more than 600Mbps domestic both UL and DL even at peak time. (1Gbps FTTH, KDDI affiliated ISP.With NTT affiliated I got much slower). Since traffic between xTom is domestic I'm expecting similar speed.

  • @reisenpai said:

    From the data of my vpn provider, 10G for 2000 people is pretty alright.
    I would say you can get at least 500Mbit most of the time, perhaps 200 in peak hour.

    I don't understand why this concerns you, just look at the price.

    Lower your expectations, you are looking at something that's 5x cheaper than average.

    I don't have high expectation but shared 1Gbps port scared me a bit. I'm not sure 1Gbps is good or bad as I've never seen this data before.

    Well. I'll have to wait and try the actual experience.

    High hopes low expectations.

  • VirMachVirMach Member, Patron Provider

    San Jose Update - A couple nodes have been 100% set up here, I'm just waiting for the templates to sync and then testing a VM and will begin deploying San Jose preorders.

    @reisenpai said: I don't understand why this concerns you, just look at the price.

    Lower your expectations, you are looking at something that's 5x cheaper than average.

    To be clear, what we are providing in a premium level of bandwidth and just because the price is low it doesn't mean it's any less. I do see a lot of providers in that area that have similar low pricing provide less bandwidth than usual but we made sure to get enough bandwidth, regardless of costs, to be able to offer the same level, even in Tokyo. Based on how much we're paying, we probably should have charged way more for these plans but I did say this will be the lowest pricing all year and I meant it.

    This is just a very good deal. It doesn't mean we cheaped out on the location in any way. Of course, and I'm sure this will be misinterpreted as well, we didn't get the best possible connection to China, because that would cost something ridiculous, like $50,000 per month for 1Gbps.

  • pddpdd Member

    we didn't get the best possible connection to China, because that would cost something ridiculous, like $50,000 per month for 1Gbps.

    You are right! Mjjs will even pay several times more than yours just for a good connection to China.Prices are generally in the tens of dollars a year

  • @VirMach said:

    @reisenpai said: I don't understand why this concerns you, just look at the price.

    Lower your expectations, you are looking at something that's 5x cheaper than average.

    To be clear, what we are providing in a premium level of bandwidth and just because the price is low it doesn't mean it's any less. I do see a lot of providers in that area that have similar low pricing provide less bandwidth than usual but we made sure to get enough bandwidth, regardless of costs, to be able to offer the same level, even in Tokyo. Based on how much we're paying, we probably should have charged way more for these plans but I did say this will be the lowest pricing all year and I meant it.

    This is just a very good deal. It doesn't mean we cheaped out on the location in any way. Of course, and I'm sure this will be misinterpreted as well, we didn't get the best possible connection to China, because that would cost something ridiculous, like $50,000 per month for 1Gbps.

    Yes, I know.

    His thoughts were very off, that's why I replied such.

    No doubt this is a very good deal.

  • yoursunnyyoursunny Member, IPv6 Advocate

    @VirMach said:

    @FrankZ said: Virmach has a average network bandwidth usage of 40 Mbps (~320 mbp/s) per node. As each node has at least a 1gbp/s uplink, the network interface each node runs on is about 33% capacity, on average.

    I know this is difficult to believe as it is for me as well, but when I said 40 Mbps I specifically meant Megabits, as in 5MB (Megabytes) per second. For all the infinite bandwidth providers hand out, only about that much seemingly gets used per node and this is on our beefy 256GB RAM + dual Xeon servers.

    Is this an open invitation for everyone to blast the ports?
    I know a few ways to egress 100 MB/s using 33% CPU.
    I just need all of LET to watch my push-ups simultaneously for 24x7.

    @VirMach said:
    That means you are more likely to be migrated than remain in Buffalo.

    Is Los Angeles to San Jose migration or Atlanta to Miami migration ever going to happen?

  • VirMachVirMach Member, Patron Provider

    @yoursunny said:

    @VirMach said:

    @FrankZ said: Virmach has a average network bandwidth usage of 40 Mbps (~320 mbp/s) per node. As each node has at least a 1gbp/s uplink, the network interface each node runs on is about 33% capacity, on average.

    I know this is difficult to believe as it is for me as well, but when I said 40 Mbps I specifically meant Megabits, as in 5MB (Megabytes) per second. For all the infinite bandwidth providers hand out, only about that much seemingly gets used per node and this is on our beefy 256GB RAM + dual Xeon servers.

    Is this an open invitation for everyone to blast the ports?
    I know a few ways to egress 100 MB/s using 33% CPU.
    I just need all of LET to watch my push-ups simultaneously for 24x7.

    @VirMach said:
    That means you are more likely to be migrated than remain in Buffalo.

    Is Los Angeles to San Jose migration or Atlanta to Miami migration ever going to happen?

    Migration to San Jose depends, we'd need to move people from LA to San Jose and then San Jose to Tokyo so it's more difficult. Los Angeles is more likely migration to Tokyo right now based on the numbers. I'm sure all of this will actively shift though and we have minimal control over it.

    Miami, I'm actually trying to work out a deal here but it's difficult. The quotes we're getting are closer to exotic locations than they are to somewhere like Tampa, Florida. The issue is the companies that do Miami, we don't have other presence with them so their first quote is going to be difficult to negotiate as well. On top of this we have to split up and redo some IPv4 space to give Miami a slice if its with a different provider.

    I'm still trying to make it work, and if we did make that work then Atlanta to Miami would be an option, yes.

    Thanked by 2FrankZ yoursunny
  • @VirMach said:

    @FrankZ said: This is not my understanding, I remember VirMach saying something about older EPYC cpus in the past, but not recently. I expect VirMach will clarify the situation on if he will retain servers in buffalo.

    Epyc doesn't really work out for us. There are some situations where there will be minor improvements in pricing if we used Epyc over Ryzen, but that all just goes toward overselling/cramming people in.

    I'm saying this very informally but any provider using Epyc either really loves "Enterprise" equipment that costs more or they really love overselling.

    I've done the math many more times than I care to admit. Like an insane amount of time has been spent theorizing every possible build and every time I look at Epyc, I almost trick myself into thinking that it may be the way to go, until I run the numbers back and realize I've been correct for a dozen times and I only thought it'd be better because of hyper-inflated numbers in, for example, the RAM category.

    Of course my statement may change at a later time, depending on how Epyc motherboards and Epyc processor pricing changes over time.

    Are you only talking about buying new or used as well? I figured most Epyc providers were getting old generations on eBay. And those motherboards have bifurcation and a shitload of PCIe lanes.

  • reisenpaireisenpai Member
    edited March 2022

    @foitin said:
    Backing up whole disk, not frequently but hope burst speed is decent.

    Even less of a concern than. A lot of upstreams doesn't even count ingress traffic of their customers.
    Yes it's the same reason you get asymmetric down/uplink from most residential connections.

    Edit: Grammar

  • @VirMach said:
    and I'm sure this will be misinterpreted as well, we didn't get the best possible connection to China, because that would cost something ridiculous, like $50,000 per month for 1Gbps.

    So what you're saying is, that they can expect as good as the best connection to China. Got it.

    Thanked by 1FrankZ
  • brueggusbrueggus Member, IPv6 Advocate

    @VirMach said: The issue is the companies that do Miami, we don't have other presence with them so their first quote is going to be difficult to negotiate as well.

    Doesn't Dedipath have a POP in Jacksonville, FL?

  • VirMachVirMach Member, Patron Provider

    @TimboJones said:

    @VirMach said:

    @FrankZ said: This is not my understanding, I remember VirMach saying something about older EPYC cpus in the past, but not recently. I expect VirMach will clarify the situation on if he will retain servers in buffalo.

    Epyc doesn't really work out for us. There are some situations where there will be minor improvements in pricing if we used Epyc over Ryzen, but that all just goes toward overselling/cramming people in.

    I'm saying this very informally but any provider using Epyc either really loves "Enterprise" equipment that costs more or they really love overselling.

    I've done the math many more times than I care to admit. Like an insane amount of time has been spent theorizing every possible build and every time I look at Epyc, I almost trick myself into thinking that it may be the way to go, until I run the numbers back and realize I've been correct for a dozen times and I only thought it'd be better because of hyper-inflated numbers in, for example, the RAM category.

    Of course my statement may change at a later time, depending on how Epyc motherboards and Epyc processor pricing changes over time.

    Are you only talking about buying new or used as well? I figured most Epyc providers were getting old generations on eBay. And those motherboards have bifurcation and a shitload of PCIe lanes.

    I'm talking about used prices for the processors, yes. Although it's a weird market in general, sometimes it'll be cheaper brand new especially for the newer ones if they have too much stock.

    Thanked by 1TimboJones
  • VirMachVirMach Member, Patron Provider

    @brueggus said:

    @VirMach said: The issue is the companies that do Miami, we don't have other presence with them so their first quote is going to be difficult to negotiate as well.

    Doesn't Dedipath have a POP in Jacksonville, FL?

    Yep, but that's not really close to Miami. It's probably closer to Atlanta than Miami geographically. I do know that it's supposed to have good connection to certain regions either way, I just really wanted Miami.

    If something goes wrong it's also a lot nicer flying into Miami.

  • Didn't know about this Miami "plan" :-)

  • @bakageta said:

    @noisycode said:
    My bad. I was wondering if the communication between two VPSes, which are both from VirMach, located in Tokyo datacenter, would suffer from 40Mbps as well.

    Nothing "suffers" from 40mbit, that's just what the average usage ends up being. I regularly transfer things between two LA VPSs and while it doesn't hit a full gigabit all that often, I don't think I've got less than 500mbit. I can't imagine Tokyo being any different. It'll count against your monthly bandwidth transfer for each server, but it shouldn't be slow.

    Cool. I was just curious about the corner case, where there is jam. To be clear, I'm OK with 2Mbps international bandwidth (inbound+outbound), while with 100Mbps at least local (intranat) bandwidth across nodes.

  • DoucheBagzDoucheBagz Member
    edited March 2022

    @VirMach is there a possibility of moving my undeployed Amsterdam service to San Jose instead? I did create a ticket regarding this, but there has been no reply for 3 days. #450016

  • @DoucheBagz said:
    @VirMach is there a possibility of moving my undeployed Amsterdam service to San Jose instead?

    You'd better ask for a refund and purchase another service in SJ, cause switching to another plan is currently not supported, according to VirMach.

  • @VirMach said:

    @brueggus said:

    @VirMach said: The issue is the companies that do Miami, we don't have other presence with them so their first quote is going to be difficult to negotiate as well.

    Doesn't Dedipath have a POP in Jacksonville, FL?

    Yep, but that's not really close to Miami. It's probably closer to Atlanta than Miami geographically. I do know that it's supposed to have good connection to certain regions either way, I just really wanted Miami.

    If something goes wrong it's also a lot nicer flying into Miami.

    How is the progress of the Tokyo computer room now?

  • @VirMach said:

    @Sherry said:

    @VirMach said:

    @cleanerleon said:
    @VirMach I just take 2 plans on same location, and does the communication between two servers consume the bandwidth quota?

    Yes, there's no internal networking setup at the time.

    View Ticket #355597
    Tickets have not been dealt with what happens? Can anyone tell me?

    Oh hey looks like you provided it. Let me have a look at why it wasn't processed. It still won't be processed just because you provided it here but I'll provide reasoning.

    View Ticket #355597
    Tickets have not been dealt with what happens? Can anyone tell me?

  • yoursunnyyoursunny Member, IPv6 Advocate

    @VirMach said:

    @yoursunny said:
    Is Los Angeles to San Jose migration or Atlanta to Miami migration ever going to happen?

    Miami, I'm actually trying to work out a deal here but it's difficult. The quotes we're getting are closer to exotic locations than they are to somewhere like Tampa, Florida. The issue is the companies that do Miami, we don't have other presence with them so their first quote is going to be difficult to negotiate as well. On top of this we have to split up and redo some IPv4 space to give Miami a slice if its with a different provider.

    What a surprise!
    I thought I would get an instant reply from @FrankZ "there's no Miami location".

    Tempa and Miami are same for me, as long as they have good connection to at least some part of South America.

  • @yoursunny said: What a surprise!
    I thought I would get an instant reply from @FrankZ "there's no Miami location".

    I'm a bit busy with other things today, but my response would have been....

    Although VirMach does not have a Miami location at this time, since nobody can truly understand the mind of VirMach there could be a Miami location in the future. Although this is not guaranteed. Please do not submit a ticket requesting a migration to the new Miami location that VirMach may or may not be thinking about.

    P.S. You seem a bit passive aggressive these days, have I done something inadvertently to piss you off?

  • julensmjulensm Member
    edited March 2022

    @VirMach

    This post has nothing to do with the thread, but since you're quite active here I decided to try to get in touch with you here.

    Could you or have someone check my ticket, please?
    Ticket #759560
    It has been more than 48 hours in 'waiting' status.

    Thank you.

This discussion has been closed.