Howdy, Stranger!

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


DACENTEC >>> Dedicated Servers starting $35/mo - 24/7/365 On Site Support
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.

DACENTEC >>> Dedicated Servers starting $35/mo - 24/7/365 On Site Support

dacentecdacentec Member, Host Rep

The Dacentec Difference

• 7 Day no questions asked money back guarantee for accounts without spam or abuse.
• 7 Day grace period for non-payment.
• No notification period for cancellations. Cancel anytime.
• No bandwidth overage charges. One warning without penalty, then a choice of upgrading service or capping bandwidth.
• No long term commitment or lock in, if your business model changes, end customer cancels or you just decide you want to do something different you can cancel all or part of your services with us.
• Flexible billing terms, pay monthly, quarterly, semi-annually or annually. Select the payment date that suits you best.
• 24 Hour setup SLA, even on the weekends, ensures you have the resources you need when you need them. Most servers are delivered within a few hours.
• Dedicated VLAN, your resources aren't sitting in a 'pool' with other customers resources.
• 24/7/365 Onsite North America Based Sales and Support Gigabit ports

All Dedicated Server Accounts
• Hardware replacement including disk is free during the rental period
• Free IP KVM loan on request
• Remote Reboot
• Automated OS Installs
• Bandwidth Graph
• Gigabit Port 10 TB monthly transfer.
• /29 IP Addresses (w/ justification) - Additional IP's Also Available For Order

Unmanaged Dell 2XL5420 16GB 2x2TB
ORDER NOW FOR $35/MONTH

Unmanaged Supermicro 6-bay Opteron 1385 8GB 2x2TB Rent Only Special (Bonus 4TB FREE) -- Limited Availability - ORDER NOW FOR $25/MONTH

Unmanaged Supermicro 4-Bay E3-1270 V3 32GB 2x2TB SATA Rent Only -
ORDER NOW FOR $60/MONTH

[>>>>>>VIEW OUR COMPLETE LIST OF SERVERS HERE<<<<<<]

SSAE16 SOC 2 Custom Rooms, Suites, Half and Full Cabinets No long term commitments – Flexible month-to-month terms Premium Blended Bandwidth Carrier Neutral – Diverse providers Dual Utility / UPS / Generator

Test IP 199.255.156.6

Test Site http://mirror.dacentec.com

Test Files http://mirror.dacentec.com/1000MB.bin http://mirror.dacentec.com/100MB.bin http://mirror.dacentec.com/10MB.bin

Comments

  • Why is your speedtest URL fronted by cloudflare? That kind of defeats the purpose of the test files.

    Thanked by 1lukesUbuntu
  • dacentecdacentec Member, Host Rep

    @hacktek said:
    Why is your speedtest URL fronted by cloudflare? That kind of defeats the purpose of the test files.

    We are working with our team to resolve this.

  • dacentecdacentec Member, Host Rep

    @hacktek said:
    Why is your speedtest URL fronted by cloudflare? That kind of defeats the purpose of the test files.

    We have resolved that issue now.

  • Nice. I've had dacentec dedis before and currently have the 8 TB Opteron. Support is top notch. Hardware is old and depending on what your network workload is you might get null routed due to old hw and unoptimized drivers so you really don't want to run heavy network workloads constantly, at least on Ubuntu. Bursting seems fine, have yet to be null routed for that but I haven't burst above 400 Mbps yet.

  • yongsikleeyongsiklee Member, Patron Provider

    "...• Gigabit Port 10 TB monthly transfer...."
    Are you sure it is on Gigabit port?
    By looking glass, the speed never reaches/goes over 100 Mbps.

  • zevuszevus Member
    edited March 2020

    @yongsiklee said:
    "...• Gigabit Port 10 TB monthly transfer...."
    Are you sure it is on Gigabit port?
    By looking glass, the speed never reaches/goes over 100 Mbps.

    iperf3 Network Speed Tests (IPv4):
    ---------------------------------
    Provider                  | Location (Link)           | Send Speed      | Recv Speed
                              |                           |                 |
    Bouygues Telecom          | Paris, FR (10G)           | 853 Mbits/sec   | 393 Mbits/sec
    Online.net                | Paris, FR (10G)           | busy            | busy
    WorldStream               | The Netherlands (10G)     | 850 Mbits/sec   | 290 Mbits/sec
    wilhelm.tel               | Hamburg, DE (10G)         | 845 Mbits/sec   | 442 Mbits/sec
    Biznet                    | Bogor, Indonesia (1G)     | 631 Mbits/sec   | 95.9 Mbits/sec
    Hostkey                   | Moscow, RU (1G)           | 793 Mbits/sec   | 778 Mbits/sec
    Velocity Online           | Tallahassee, FL, US (10G) | 917 Mbits/sec   | 412 Mbits/sec
    Airstream Communications  | Eau Claire, WI, US (10G)  | busy            | 877 Mbits/sec
    Hurricane Electric        | Fremont, CA, US (10G)     | 896 Mbits/sec   | 338 Mbits/sec

    I've had about 10 diff servers with Dacentec over last 6 years or so. I am a bit disappointed in that their hardware has not changed much and theier price point is only slightly lower, they used to be one of the top budget dedi providers ... but they are still 'relatively' cheap and reliable -- outside of this October 2019 incident, heh .. I remember my brother being quite angry about it, think he was down for some 8 hours. This is not the norm. I'd say my uptime over the years is about 99.995%.

    I use one of the $20 servers now -- think network speeds would be a bit higher, but my HDD (and CPU) bottlenecks to some extent. I was also running bitcoind when I did the test.

  • dacentecdacentec Member, Host Rep

    @hacktek said:
    Nice. I've had dacentec dedis before and currently have the 8 TB Opteron. Support is top notch. Hardware is old and depending on what your network workload is you might get null routed due to old hw and unoptimized drivers so you really don't want to run heavy network workloads constantly, at least on Ubuntu. Bursting seems fine, have yet to be null routed for that but I haven't burst above 400 Mbps yet.

    Thanks for the comments on our support team. Client satisfaction is something we take a lot of pride in and we do our best to make hosting as uncomplicated as possible for our clients. We do have newer servers hitting the site for order, including E5's. The most common cause of self-induced null routes are misconfigured automated backups generating minor packet floods directed at an IP akin to a minor DoS attack and triggering the Data Center's internal automated DDoS mitigation system. Once the backup process is halted and properly reconfigured, the system will automatically remove the existing null route. Typically, our support team can help identify if this is a potential cause of a null route and advise.

  • @dacentec said:

    You have a great service that I've also used before. However, the above comments regarding null routing and network drivers are accurate and are a big issue with your service.

    For some reason, the default network drivers in use on many of your servers tend to generate packet floods when passing lots of traffic. This means common utilities like rsync generate far higher packet rates than they do on other modern servers. So you have users being null routed for abuse who have no intention or desire to cause trouble; they're just doing network copies or backups as they're accustomed to elsewhere.

    If possible, you might try to help users to find and install updated network drivers for certain common OSes as a way of solving this problem. I know you provide an "unmanaged" service, but this is a case where it helps everybody if you took a slightly more active role in helping users optimize their network use. Again, I don't know if that's reasonable (or if updated drivers even exist), but it could be worth looking into if you haven't already.

  • hacktekhacktek Member
    edited March 2020

    I will say however that for my current server and my current use case (nginx caching of 16 to 32 MB chunks), I burst on the ingress to over 800 Mbps for about 30-40 minutes on saturday and the service was not null routed so it looks like not all use cases have this problem. The 8TB server has been working pretty well for this use case as well. Since this is volatile data I can live without I just did a simple mergerfs overlay of all 6 disks and I'm filling them using a least used space policy so they fill in tandem and performance is not bad. Obviously the HDDs are old but they are speedy enough to saturate bandwidth if it got to that (and the null route system allowed it): ~115-125 MB/s.

    EDIT: here's the graph...

    Note that it was a spiky increase, the ingress wasn't pegged at 820 Mbps, if it had been I'm sure that would have triggered null routing.

  • @hacktek said:

    Note that it was a spiky increase, the ingress wasn't pegged at 820 Mbps, if it had been I'm sure that would have triggered null routing.

    Possibly, but it seems to be the packet rate that they care about the most. You can check that with a utility such as iptraf. I can only say that my old Dacentec hardware was generating about 10 times as many packets as modern hardware at the same data rate. I seem to remember that the rate causing problems was somewhere in the neighborhood of 80k packets per second.

    I'm not very knowledgeable about how packet sizes are determined, and whether this is up to the network driver alone or if system-level utilities can influence it. But any high date rate transfers were a problem for me; I always had to limit bandwidth in the software to prevent a "packet flood."

    Thanked by 1hacktek
Sign In or Register to comment.