Howdy, Stranger!

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

Advertise on LowEndTalk.com
Vultr vs Virmach for cheap hosting
New on LowEndTalk? Please Register and read our Community Rules.

Vultr vs Virmach for cheap hosting

Hello LowEndTalk!

I recently came across the 10 dollar per year deals from Virmach. I already have a 3.50 dollar/month VPS with Vultr which is perfectly fine but because I use it as an occasional webserver, the Virmach plan is cheaper and has better specs with 768mb of ram and 20gigs of storage compared to the 10gig of disk space with Vultr

On one hand, I get a better dashboard, better support for more money. But with Virmach, I get better specs for less money.

What do you think is a better choice? My primary use case is just hosting a webserver and a filehost to share and host files.

Comments

  • BertieBertie Member

    @PerryPal said: But with Virmach, I get better specs for less money.

    Historically, Virmach has had much more restrictive limitations on CPU, I/O, network usage, and a very aggressive anti-abuse detection system prone to false positives. I believe they're toned it down and worked out most of the issues in the past year, but I'm not the one to test the limits. Vultr is probably the clear winner as to which provider would let you utilize their services to the fullest.

    That being said, for a simple webserver and file-host, I think you'd be fine with Virmach. As always, if it's a high availability service, plan for what should happen when it goes down and keep backups.

    Thanked by 1kkrajk
  • Go for Virmach. If availibility is really important, you probably want more than just a single Vultr VM. CDN, IP failover, etc.

    Anyway, you will get close to 100% uptime with Virmach from my experience.

  • @ben47955 said:
    Go for Virmach. If availibility is really important, you probably want more than just a single Vultr VM. CDN, IP failover, etc.

    Anyway, you will get close to 100% uptime with Virmach from my experience.

    Ah thanks. Thats all I need. I will be making offsite backups.

  • @Bertie said:

    @PerryPal said: But with Virmach, I get better specs for less money.

    Historically, Virmach has had much more restrictive limitations on CPU, I/O, network usage, and a very aggressive anti-abuse detection system prone to false positives.

    What do they do when you exceed limits?

  • @PerryPal said: What do they do when you exceed limits?

    Usually, step are
    1. They open ticket asking to reduce your usage.
    2. Shutdown VPS
    3. Suspend VPS
    4. Terminate

    If you don't act like a dick, thing don't go further point 2.

  • @ben47955 said:

    @PerryPal said: What do they do when you exceed limits?

    Usually, step are
    1. They open ticket asking to reduce your usage.
    2. Shutdown VPS
    3. Suspend VPS
    4. Terminate

    If you don't act like a dick, thing don't go further point 2.

    Thanks for your answers, just currious about maximum 50% CPU usage for 15 min. As far as I know (by currently owning 4 vps with them), their VPS already have CPU reduced to around 25-30% of normal CPU, so why I can't use all of CPUs or at least 50% all the time? Some metric applications required a little bit heavy CPUs and may easily pass 50% threashold in some high load situation.

  • @ben47955 said:
    If you don't act like a dick, thing don't go further point 2.

    Oh that sounds pretty reasonable. Thanks Virmach seems perfect then

  • Vultr feels much reliable than Virmach, but Virmach gives much more ram.

    Don't give me hope.

  • @elliotc said:
    Vultr feels much reliable than Virmach, but Virmach gives much more ram.

    This is my VPS with the worst uptime. I'm paying 0.95/y for that. Show me Vultr with better.

    Thanked by 1merlinvn
  • @ben47955 said:

    @elliotc said:
    Vultr feels much reliable than Virmach, but Virmach gives much more ram.

    This is my VPS with the worst uptime. I'm paying 0.95/y for that. Show me Vultr with better.

    Google have downtime last month, virmach is more reliable than google now.
    Reliable is not about uptime but... Reliable. Virmach feels more like a workshop, and vultr is a corporation.

    Thanked by 1kkrajk

    Don't give me hope.

  • waynechriswaynechris Member
    edited January 6

    If you don’t care about the operation quality of the website and just want to put some pages, please choose virmach,
    If you are running a major business website, you care about the quality of the operation, and don’t care about money, please choose vultr
    I have used both of these two vps. Vultr's support ticket response is relatively fast, vps migration is convenient, and virmach is like a lighter that is often lost by me. It is cheap and a necessities of life. Based on this price, don't ask too much high

    Thanked by 1merlinvn

    My mini—box
    ClouveoAMD SSD Cloud Vps & Hosting
    Clouveo Asia Regional Sales Manager[FAKE]

  • Daniel15Daniel15 Member
    edited January 6

    I've found VirMach to be fast and reliable. Vultr's CPU can actually end up slower than VirMach's unless you get their "high frequency compute" VPSes (which start at $6/month, not $3.50).

    @PerryPal said: What do they do when you exceed limits?

    They have an automated system that opens a ticket, and I think you get several warnings before they do anything drastic like throttle or shut down your VPS. I think their limits are quite reasonable these days... I doubt they'd flag anything if you use a lot of CPU for just 20 minutes or so. I have daily backups which do a lot of disk I/O and CPU usage (compressing, encrypting, and deduping data with Borgbackup) every night and I've never been flagged by their system.

    Thanked by 1merlinvn
  • @elliotc said: Google have downtime last month, virmach is more reliable than google now.

    Reliable is not about uptime but... Reliable. Virmach feels more like a workshop, and vultr is a corporation.

    What is you definition of reliability ? Availability maybe ? Like I said previously, if you want high availability you will need a more complex architecture than a single vps and vps provider would not matter that much considering I'm implying at least two different provider.

    Virmach give you enough bang for your buck and that should be enough for home project.

    Thanked by 1Erisa
  • @ben47955 said:

    @elliotc said: Google have downtime last month, virmach is more reliable than google now.

    Reliable is not about uptime but... Reliable. Virmach feels more like a workshop, and vultr is a corporation.

    What is you definition of reliability ? Availability maybe ? Like I said previously, if you want high availability you will need a more complex architecture than a single vps and vps provider would not matter that much considering I'm implying at least two different provider.

    Virmach give you enough bang for your buck and that should be enough for home project.

    I am not saying that virmach is bad, I just want to point out that they have different market niches.

    Thanked by 1Bertie

    Don't give me hope.

  • @elliotc said: I am not saying that virmach is bad, I just want to point out that they have different market niches.

    Of course, different pricing too. But I would be willing to include Virmach as Vultr in any complex architecture.

  • ariq01ariq01 Member

    Of course @VirMach dude.

    wryyyyyyyyyyyyyyyyyyyyyyyy

  • BertieBertie Member
    edited January 6

    @ben47955 said: This is my VPS with the worst uptime. I'm paying 0.95/y for that. Show me Vultr with better.

    Uptime is just one aspect of reliability. I have over a dozen Virmach services and at any given moment, at least one service is either delinked from my panel, DHCP is broken, HTML5 VNC doesn't work, a migration from OVZ is botched, etc.

    Virmach is great for the price you pay and I'm certainly not complaining that my $5 VPS has some minor difficulties in usage, but it's not really the most reliable by a longshot. This is compounded by Virmach threatening to levy $30 invoices if you submit a ticket for something that "we're aware of and handling".

    But that service is worth $10/year for a reason, and you get $10's worth of service. Not much more, not much less. Answering a single ticket might cost them $5 worth of manpower, negating any profit from that service.

    OP makes it sound like they're okay with a slightly less reliable service for a service that's a quarter of the price.

    Thanked by 1pluush
  • @Bertie said:

    @ben47955 said: This is my VPS with the worst uptime. I'm paying 0.95/y for that. Show me Vultr with better.

    Uptime is just one aspect of reliability. I have over a dozen Virmach services and at any given moment, at least one service is either delinked from my panel, DHCP is broken, HTML5 VNC doesn't work, a migration from OVZ is botched, etc.

    Virmach is great for the price you pay and I'm certainly not complaining that my $5 VPS has some minor difficulties in usage, but it's not really the most reliable by a longshot. This is compounded by Virmach threatening to levy $30 invoices if you submit a ticket for something that "we're aware of and handling".

    You call the worst of Virmach, that a bit unfair. You probably don't keep a dozen vps at Vultr to do any comparaison, but they have their issues. About ticket, it's not 30$ but 15$. I don't think they are that dick about that. They can't expect everyone to keep reading a thread of many pages for many week of course, but checking the status page before creating ticket seem legit. They uses strong word, but they are reasonable about that.

  • BertieBertie Member
    edited January 7

    @ben47955 said: You call the worst of Virmach, that a bit unfair.

    It's unfair to say that I have small issues with Virmach regularly? I'm not sure my experiences/issues at Virmach are invalidated because I don't idle 12 VMs at Vultr.

    Virmach has problems. They're not major problems and it's not a bad provider. They have great deals and I'm happy to purchase and host non-critical things with them.

    Virmach is a great company that fulfills a niche in the budget VM market, but fanboying over Virmach and being offended when your favourite provider isn't called perfect isn't really productive.

    Vultr on the other hand, fulfills a niche in the budget/mid-range cloud infrastructure market, and caters to businesses. You pay more, and you expect more from Vultr as a result. Vultr is not dramatically better than Virmach, but in my opinion, it is better overall (while Virmach might be better in terms of cost/service ratio).

    Thanked by 2pluush willK
  • If you want to test some code or let your VPS idle, VirMach is fine. If not, go for Vultr. Moreover, read this thread carefully before going for VirMach. It seems there is very limited support, even months without responses if there are issues:

    https://www.lowendtalk.com/discussion/169065/virmach-since-you-could-not-solve-the-problem-could-you-refund-the-order/p1

    Thanked by 2Bertie pluush
  • HxxxHxxx Member
    edited January 7

    Wanted to jump in and say that...
    1- You can't simply compare Virmach to a Titan like VULTR. Simply you can't. VULTR is at the level of (or even greater than) Linode and DigitalOcean. VULTR parent company is a datacenter (Choopa). The comparison of Virmach vs VULTR is so wrong at any level, is like comparing AWS with Bluehost, VULTR been the AWS in the example.

    2- You compare companies by market target. AWS is usually compared with Google Cloud and Azure. VULTR is compared with Linode and DO. VULTR targets developers same as Linode and DO. Provides a highly custom (in-house developed) control panel with Cloud like features.

    You could compare Virmach with Racknerd for instance.
    Personally and with big respects to these two brands (Virmach and Racknerd)... I wouldn't use them for anything production/business/serious.

    If you means business, and you are running critical stuff skip ahead and start with VULTR, DO, Linode. You could also consider other providers like RamNode (which like Virmach started small but now are in a whole other plane) and also Frantech / BuyVM which is one of the few with an in-house control panel and infrastructure (Francisco is kind of a genius).

    I would even say if your budget is bigger do the jump to the big boys: AWS / Microsoft AZURE / Google Cloud. Usually when you are running something critical, mentioning any of these three in a conversation with partners gives you advantage, is the "proven solution" at corporates.

    Some of you guys probably won't understand and is ok since most here are just hungry for cheapo unusable services. Then when something goes to shit as usual... threads with "my VPS at x shitty provider is down, I'm paying 2 dollars a year and I'm losing MILLLLLLIONSSSS" start to appear...

    2cents

  • BertieBertie Member
    edited January 7

    @OliverScott said: Moreover, read this thread carefully before going for VirMach. It seems there is very limited support, even months without responses if there are issues:

    Yikes, some of these responses are quite arrogant. I was open to giving Virmach the benefit of the doubt, but I guess that's something to think on.

    Thinking about it, it's quite silly to prevent users from changing the email address on the account as well unless they "paid for a support package".

    Thanked by 1pluush
  • miulmiul Member

    vu has a better line, there will be a better network experience. The CPU performance of vu is higher, while the vir limits the performance of the cpu more. vu's hard disk read and write speeds are generally better. In general, vir is more suitable for novices to practice. If you want better service quality ,vu is recommended.

  • Virmach always says that they are pure SSD disks, but their TOS does require that the average IOPS in any two hours cannot exceed 80. This threshold is lower than ordinary HDD hard disks. If you want to carve your data on a stone , Please choose VirMach.

    Thanked by 2Bertie levnode
  • VirMachVirMach Member, Provider, Top Provider

    @abiaolaoge said: Virmach always says that they are pure SSD disks, but their TOS does require that the average IOPS in any two hours cannot exceed 80. This threshold is lower than ordinary HDD hard disks. If you want to carve your data on a stone , Please choose VirMach.

    We're going to update that with better verbiage/figures. There's no case in the last 5 years where someone has actually been suspended or even powered off for 80 IOPS. It scales based on your package size, and the minimum is significantly higher than 80 IOPS. This figure was from a long time ago, when we also offered HDD services and that was for a brief period of time.

    It is meant to be there as a bare minimum where we can begin looking into taking action, but I can understand why there's concern.

    @OliverScott said: It seems there is very limited support, even months without responses if there are issues

    In the 30 days following our Black Friday offer, about 75% of people received relatively normal response times, and 25% received delayed responses. The #1 longest wait right now are those tickets regarding billing as we had issues with Alipay, that we automatically resolved and there are still many of these requests open. For billing tickets.

    Special offers are sorted differently, and they generally receive responses after our normal support level packages (unless it's an emergency situation.)

    @Bertie said: have over a dozen Virmach services and at any given moment, at least one service is either delinked from my panel, DHCP is broken, HTML5 VNC doesn't work, a migration from OVZ is botched, etc.

    • About 0.6% of services faced some "delinking" issue.
    • HTML5 VNC was broken for some time after SolusVM update. We pushed an update to all servers, but about 3% of nodes are still experiencing some level of web VNC issue. Services are still accessible via any other VNC application in these cases.
    • Network issues are rare. We did have a recent issue with nullroutes from the datacenter. This, unfortunately, did incorrectly affect 0.005% of customers (where they received a nullrouted IP on their new service.)
    • Migration from OVZ to KVM was definitely not perfect. Major issues were extremely rare, but more prominent than we had hoped or seen in testing.

    @Daniel15 said: I've found VirMach to be fast and reliable. Vultr's CPU can actually end up slower than VirMach's unless you get their "high frequency compute" VPSes (which start at $6/month, not $3.50).

    We understand that our hardware is getting outdated and working toward using processors that do end up also being better than any advertised benchmark I see from Vultr. For the bulk of special offers and potentially KVM Lite services though, we might not end up going with the same brand new hardware, we'll see.

    These are the currently planned future nodes:

    128GB 2666MHz DDR4 RAM, Ryzen 9 3950X @ 3.5GHz, 4x2TB NVMe (RAID 10.)

    @ben47955 said: Usually, step are
    1. They open ticket asking to reduce your usage.
    2. Shutdown VPS
    3. Suspend VPS
    4. Terminate

    If you don't act like a dick, thing don't go further point 2.

    @Bertie said: Historically, Virmach has had much more restrictive limitations on CPU, I/O, network usage, and a very aggressive anti-abuse detection system prone to false positives. I believe they're toned it down and worked out most of the issues in the past year, but I'm not the one to test the limits.

    Currently, our anti-abuse system related to I/O and CPU does the following:

    1) Ignores usage in certain cases completely.
    2) Throws out the first case if it's below a certain threshold.
    3) Provides a light warning.
    4) Potentially provides a second warning.
    5) Provides another warning, but this time with a powerdown.
    6) May repeat the step above any number of times to avoid going to the next step.
    7) Suspends the service in extreme cases (almost never, mostly disabled.)
    8) No termination occurs, outside of customer cancelling or not renewing.

    Nearly all cases of CPU related power-downs are with the customer being at 99% or more CPU usage for at least 2 hours. Average I/O related powerdowns seem to be around 5,000 operations per second over many hours, although it ranges from around 200 to 80,000 (with some outliers here and there.)

    Once we roll out the Ryzen 3950X packages, we'll basically have a 4x improved CPU to RAM ratio so we can potentially allow people to essentially have "dedicated" CPU on those packages, but we'll see. And since it will be NVMe's we can be much more lenient on I/O as well although I'd like to think we already are pretty lenient. Everyone gets to use multiple times more I/O over something like 6-8 hours than their share of disk space on the node. This means if the sevice is 100GB disk and the node has 8x1TB SSDs in RAID10, you could basically burst to 3-5x 100GB/4000GB, or as much as something like 12.5% of all available estimated disk operations for your 2.5% share of the disk.

    @Bertie said: Thinking about it, it's quite silly to prevent users from changing the email address on the account as well unless they "paid for a support package".

    We just had to do this, I'm sorry.

    For whatever reason, the percentage of customers requesting email changes is astronomically higher on special-only accounts. We received a thousand plus of these requests, just to put it in perspective, compared to maybe 50 of them over the entire year otherwise from standard accounts. There are also other security concerns involved. It was well thought out and it's the only solution we can implement until we code something smarter to deal with them outside of letting a customer change it automatically with little to no requirements.


    I hope that's overall some helpful information.

  • bobebobe Member

    Hello VIRMACH

    My VPS has an error in running a necessary program. Two similar errors in the past 20 days have caused me heavy losses. Yesterday, a professional checked and found that the error was:
    [2021-01-06 06:56:51.152] [error] [main] AES-256-GCM is not available on this CPU

    the CPU in that machine is very old and doesnt support encryption used

    cat /proc/cpuinfo is model : 13 AND model name: QEMU Virtual CPU version (cpu64-rhel6)

    Confirmed, no aes in flags,so I need to change the server or throw away

    Would you pls help me to deal this VPS

    Palyafc

  • swat4swat4 Member

    @bobe said:
    Hello VIRMACH

    My VPS has an error in running a necessary program. Two similar errors in the past 20 days have caused me heavy losses. Yesterday, a professional checked and found that the error was:
    [2021-01-06 06:56:51.152] [error] [main] AES-256-GCM is not available on this CPU

    the CPU in that machine is very old and doesnt support encryption used

    cat /proc/cpuinfo is model : 13 AND model name: QEMU Virtual CPU version (cpu64-rhel6)

    Confirmed, no aes in flags,so I need to change the server or throw away

    Would you pls help me to deal this VPS

    Palyafc

    :| This is not a helpdesk thread.

    Looking forward to a wider collection of VPS and domain names.

  • @bobe said:
    Hello VIRMACH

    My VPS has an error in running a necessary program. Two similar errors in the past 20 days have caused me heavy losses. Yesterday, a professional checked and found that the error was:
    [2021-01-06 06:56:51.152] [error] [main] AES-256-GCM is not available on this CPU

    the CPU in that machine is very old and doesnt support encryption used

    cat /proc/cpuinfo is model : 13 AND model name: QEMU Virtual CPU version (cpu64-rhel6)

    Confirmed, no aes in flags,so I need to change the server or throw away

    Would you pls help me to deal this VPS

    Palyafc

    If you are running shadowsocks, I recommend using chacha20-ietf-poly1305 encryption algorithm for better compatibility

  • bobebobe Member

    Thank you!

    I am not using Shadowsocks

  • _MS__MS_ Member

    @VirMach said: These are the currently planned future nodes:

    128GB 2666MHz DDR4 RAM, Ryzen 9 3950X @ 3.5GHz, 4x2TB NVMe (RAID 10.)

    Which location(s)?

  • JabJabJabJab Member

    @MS said: Which location(s)?

    From other topic:

    Psychz for Los Angeles, and we're looking to also do Amsterdam right after Los Angeles since that's in high demand.

    Thanked by 1_MS_
  • BertieBertie Member

    @VirMach said: About 0.6% of services faced some "delinking" issue. HTML5 VNC was broken for some time after SolusVM update. We pushed an update to all servers, but about 3% of nodes are still experiencing some level of web VNC issue. Services are still accessible via any other VNC application in these cases. Network issues are rare. We did have a recent issue with nullroutes from the datacenter. This, unfortunately, did incorrectly affect 0.005% of customers (where they received a nullrouted IP on their new service.) Migration from OVZ to KVM was definitely not perfect. Major issues were extremely rare, but more prominent than we had hoped or seen in testing.

    If you're answering questions - I have one that I'll never be able to submit a ticket about because ISO installation is a possibility:

    Why are root passwords non-functional 80% of the time when installing certain Debian KVM templates?

  • verovero Member, Provider

    @Hxxx said:
    Wanted to jump in and say that...
    1- You can't simply compare Virmach to a Titan like VULTR. Simply you can't.

    That's it, wonder why nobody else noticed that fact. Still, being compared to something greater is cool advertising already.

    Some of you guys probably won't understand and is ok since most here are just hungry for cheapo unusable services. Then when something goes to shit as usual... threads with "my VPS at x shitty provider is down, I'm paying 2 dollars a year and I'm losing MILLLLLLIONSSSS" start to appear...

    This is the funny part about LET IMO. But it has some charm.. Guys share knowledge and they quickly figure out who is who here.

  • Daniel15Daniel15 Member
    edited January 7

    @bobe said: cat /proc/cpuinfo is model : 13 AND model name: QEMU Virtual CPU version (cpu64-rhel6)

    Confirmed, no aes in flags,so I need to change the server or throw away

    Open a ticket and ask them to enable CPU host passthrough. Then you'll see the real CPU model, and AES-NI will be available for use. The CPU is most likely a Xeon E3 or E5.

  • BertieBertie Member
    edited January 7

    @Daniel15 said: Open a ticket and ask them to enable CPU host passthrough. Then you'll see the real CPU model, and AES-NI will be available for use. The CPU is most likely a Xeon E3 or E5.

    Alternatively, I believe the (mostly undocumented) work-around is to install an OS from the template, followed by the ISO install. If you install directly from ISO, the VM doesn't have CPU flags enabled.

  • @Bertie said:

    @Daniel15 said: Open a ticket and ask them to enable CPU host passthrough. Then you'll see the real CPU model, and AES-NI will be available for use. The CPU is most likely a Xeon E3 or E5.

    Alternatively, I believe the (mostly undocumented) work-around is to install an OS from the template, followed by the ISO install. If you install directly from ISO, the VM doesn't have CPU flags enabled.

    The trick is reinstall to templates with "gen2", debian 8 personally.
    Then, you can reinstall to any other os you want.

    Thanked by 2Bertie abiaolaoge

    Don't give me hope.

  • mirocmiroc Member

    Without this whole 'service not provisioned in time / 9+ day ticket response time' shenanigans, I would choose anyone over VirMach only because of their strict resource limitations.

    It's like "Hey, take these 3 cpu cores, BUT ACTUALLY please don't use them, we don't do that here"

    Thanked by 1pluush

    NexusBytes is epic B) | check them out! (aff.)

  • BertieBertie Member
    edited January 7

    @elliotc said: The trick is reinstall to templates with "gen2", debian 8 personally.

    Then, you can reinstall to any other os you want.

    Thanks for being more specific. I've personally just reinstalled with Debian 9 and that's been fine too.

    @miroc said: Without this whole 'service not provisioned in time / 9+ day ticket response time' shenanigans, I would choose anyone over VirMach only because of their strict resource limitations.

    Honestly, I used to think similarly. I held off on BF 2019 and only picked up 2 services, and they worked well enough at the time that I picked up much more during BF 2020. One of my VMs had 3 cores so I just CPU limited to 1 core and that seemed to work out fine with no complaints from Virmach.

    They're perfectly fine for personal use. I don't have any resentment towards Virmach, but I find it super unfortunate when any criticism of Virmach is shut down because Amir threatens not to hold more sales or "my service has been perfect, you're lying". Virmach can't be compared to Vultr, but that doesn't stop people from claiming Virmach is on par with large IaaS providers.

    Virmach is certainly not perfect, and holding hosts on a pedestal (thinking they're perfect) is really counter-productive to what a hosting forum should be doing: offering a place to candidly share feedback and advice about hosting.

    I understand that Virmach is flooded with tickets. I try and avoid opening them myself unless my service is literally unusuable. I get that it's a low-margin service and that I'm not entitled to open 3 tickets and demand an urgent response for the low prices I'm paying, but support is still very lacking and Virmach is a poor option in any case where you're doing something important/production with your service. Took me days to get an unusable service addressed after "awaiting technical review".

    I didn't care that much because it wasn't like I was hosting a web store on it or anything; just doing my thing in the meantime.

  • VirMachVirMach Member, Provider, Top Provider

    @Bertie said: Why are root passwords non-functional 80% of the time when installing certain Debian KVM templates?

    It depends on the node you're on and the specific operating system/kernel version. At some point in time SolusVM's functionality broke or some template-related operations. It's on the list to try to get it uniform and re-do these templates that we have direct from SolusVM in a way where it works with the reset features. We did correct most of these already, for example, Debian 9 had this issue and it was resolved, unless a new SolusVM update broke the functionality in some way. We encourage you to report these to us because this is how we can identify and resolve it more quickly for the exact version on the exact node your service is on.

    It will always be possible to replicate these buttons to some degree by essentially doing what SolusVM tries to do. If these functions do not work, you can also contact us and as long as you clearly state the issue is with the specific control and how it behaves, we can do the function for you.

    We do also have some helpful selectable ISOs that should assist you with password resets and partitioning.

    @Bertie said: I have one that I'll never be able to submit a ticket about because ISO installation is a possibility

    If you're reporting an issue with an image, and include all the details for us to identify and resolve the issue, feel free to make a ticket.

    But just don't expect it to be resolved for immediate usage, it would take time and just be added to our list.

    @miroc said: I would choose anyone over VirMach only because of their strict resource limitations.

    It's like "Hey, take these 3 cpu cores, BUT ACTUALLY please don't use them, we don't do that here"

    If everyone was provided completely dedicated resources on virtual servers, the biggest part of their appeal would be lost for most people. We do have "VDS" services, and we have dedicated servers if that's the route you should want or need to take, but they are priced accordingly.

    As for bursting usage, it's completely fine and in most cases you can usually use it as if it's dedicated resources as long as your service itself is not essentially struggling to keep up with the resources it has allocated. Allow me to share all of today's warnings and powerdowns with you. Maybe this will paint a clearer picture.

    • Two services (SSD256 and KVM-Special-512) powered down for CPU usage, after being warned 7-9 days ago. Average 99.97 to 107.54% CPU (one core) over a period of 2 hours or more. Re-occurred 3 times and in many of these cases continued this way for the entire week.
    • Twenty-one services powered down for I/O usage. Consists of 14x SSD256, 1x OV-Special-384, 1x SSD512, 1x Pro+ Lite, 1x BF-Special-2020, 1x BF-Special 2018, 1x SSD1G, and 1x SSD2G. Thirteen of these were SSD256's ordered by the same person across multiple accounts (usually mass abuse.)

    For the I/O abuse above, the lowest usage out of all twenty-one of them, on an SSD256, was 431.15 transfers per second (number of I/O requests) with previous usages of 1,291.26 and 1,866.37 -- this is over a period of 8 hours. The average over a period of 24 hours was 444.53 TPS, 3105.85 read requests per second, 278.79 write requests per second, at an average data size of 3973.12 bytes read per second and 168.96 bytes write per second.

    The highest usage was also on an SSD256. This one resulted in an immediate shutdown, as in an emergency shutdown with no prior warning. After the shutdown, the CPU usage on the node dropped from 66% to 51% (on a 32 core server) and I/O usage dropped from about 50% (of 8x1TB SSDs in HW RAID 10) to 10-15% so this is an example of our system taking action to improve the quality of service for everyone on that node as well. By emergency shutdown, it does not mean it gives the customer a few minutes and then shuts it down -- this only kicks in if it's also the result of overloading the server to an extreme level (and is super rare.) This one happened for a few hours, sustained, until the system intervened, which gives plenty of time for users to burst should they be installing a software or other temporarily intensive tasks. This was an average of 53,178.94 TPS with 425,032.16 read and 265.79 write requests per second. This was also around 4,000 bytes read per second and causing the device to reach near saturation levels.

    Let me know if you have any suggestions for further improving the system based on the figures above, or if you have any more specific questions.

    @Bertie said: I understand that Virmach is flooded with tickets. I try and avoid opening them myself unless my service is literally unusuable. I get that it's a low-margin service and that I'm not entitled to open 3 tickets and demand an urgent response for the low prices I'm paying, but support is still very lacking and Virmach is a poor option in any case where you're doing something important/production with your service. Took me days to get an unusable service addressed after "awaiting technical review".

    Even for limited support packages, response times will drastically improve after the next few days as we are almost caught up on the backlog of tickets.

    We've sorted them out to where it's much faster to complete them now and should complete somewhere around 800 to 1,000 tickets in one day today. Most of these tickets are already resolved, via automation. We're just giving the final responses/closing it out and letting everyone know it was already resolved.

  • pluushpluush Member
    edited January 10

    The threat of asking you to pay at VirMach, if you even slightly overuse one of your VPS (which you may not always monitor all the time, and let's say you might have gotten a DDOS on a non DDOS protected IP) made me end all my services in VirMach. You can pay $1.25 a month for a service, and a $25 surprise bill might even appear in a sudden. To me, even properly using VirMach VPSes worries me. At one time I had 10-20 VPSes on VirMach. In Vultr I had a bad code running that were using 100% of CPU for days (on one VPS) without even being warned by Vultr. After I found out about it on their panel I immediately fixed the issue, and my CPU usage on other VPSes were relatively low.

    Sometimes I feel like processing power should also be pooled and averaged from all your VPSes just like bandwidth in Linode.

    These were my experiences with VirMach, I'll stick with Vultr for core infrastructures in my project.

    $25 may not sound like much especially by U.S. standards, but for example in Indonesia if you're using the minimum wage set by the government in the capital it's roughly 2 full days worth of work.

  • VirMachVirMach Member, Provider, Top Provider

    @pluush said: and a $25 surprise bill might even appear in a sudden.

    We haven't added $25 suspension/administration fees outside of automated chargebacks for probably a year. No one really paid those nor did we force people to pay them or send them to collections, it's just there to indicate the account was not in good standing but we shifted to mainly using another method to block future orders until previous issues are resolved first.

    @pluush said: In Vultr I had a bad code running that were using 100% of CPU for days (on one VPS) without even being warned by Vultr

    We don't see this as a positive. If we allow someone to max out CPU, which does usually occur as a result of errors, we prefer, and most customers prefer it's powered off. Most customers realize this, fix their code, and let us know it's resolved. It doesn't really end how you are imagining it.

    The only difference is while on Vultr it took you days as you mentioned without even being warned until you happened to notice it, we would send a warning without power-down and then a warning with power-down should you miss the previous communication and the abnormal usage continues. These are what most cases are these days, the customer using the warning to correct the issue, outside of a few mass abusers that are clearly signing up for many accounts and maxing out usage on purpose, over and over, until they're suspended.

    @pluush said: Sometimes I feel like processing power should also be pooled and averaged from all your VPSes just like bandwidth in Linode.

    Generally, if you require a certain amount of processing power, we would recommend you go with packages that suit your needs instead if you plan on bursting them at the same time. The bandwidth idea though is interesting/feasible and perhaps if we can in the future we will do that as an option.

    Thanks for the feedback.

    Thanked by 1TimboJones
  • Honestly i've choosed Virmach for:

    • prices when discounted
    • lot of public communications like you can see here
    • acceptable / common AUP
    • good uptime

    But i would not put critical prod on Virmach cause of this:

    • relatively weak support (you have to go straight to the point / investigate for them and then they will do the job, else they will just not check / try to understand what you are saying)
    • support answers quality are far away from what we can see publicly on LET
    • some panel features are buggy or could be more ergonomic
    • i fear when i read parts of terms with content like "if you do that we may terminate your service because we don't want you as a customer" even if i don't host anything heavy, dangerous or not allowed by law
    • owner may feel like rude / in powertrip sometime
    • maybe related to "best effort support" when you have special offers but not possible to open multiple tickets to split different subjects you may have to talk about (iirc i had an automatic ticket to change rdns and i was not able to open a ticket about a panel issue)

    I don't like much when a provider makes me feel like a bad customer or a kid ^^. But it's their business and their choices and i've accepted to deal with it.

    I'm a little customer with non critical stuffs so i go with known providers on LET with good offers usually ; and i don't need support if it's not an issue on provider's side. So i was aware about what i dislike regarding Virmach but totally open to deal with it as i don't need much support, providers like Virmach do their best to offer great uptimes and fix technical issues quick.

    Regarding your needs, it looks a bit like mine, so probably Virmach is a good fit.

    I would like to suggest you to take a look also on NexusBytes (owned by @seriesn) and BuyVM (owned by @Francisco) ; these guys are always willing to help and makes you feel more than welcome. Their terms and AUP are btw super clear. Francisco is by far an awesome tech guy supported by an active community. UUnfortunately both of them doesnt offer discounts.

    DevOps and Site Reliability Engineer. Looks cool when i know what i do.
    Doing useless stuffs on amazing providers services because... Why not?
Sign In or Register to comment.