The Webdock Performance Guarantee
Last updated: October 20th 2020
We claim we have great performance compared to other hosts, and indeed we do. We have always aimed at providing some - if not the best - performance for money in the industry.
Now, competing on performance turns out is not really that hard. In this article we will explain what we mean by our Performance Guarantee, how we intend to uphold that guarantee and why we are able to provide more compute, memory and bandwidth than the competition - in addition to our fine-tuned Perfect Server Stacks.
The Webdock Performance Guarantee
We are not afraid to issue the following guarantees:
- If you find a VPS provider where after running a YABS benchmark you can show performance numbers that are higher than ours and whose pricing is lower or equal than ours. Well, at that point we would probably consider lowering our prices and/or increase whatever resource is falling behind for all customers on the platform.
If we judge it to be more of an "edge case" we will offer you an appropriate discount on your Webdock server.
Please note: When running YABS on Webdock servers, inaccurate numbers will be reported for disk performance for Micro and Micro+ instances as disk space is not sufficient to run the test properly. Please see our tickets here.
Network connectivity/performance is excluded from this guarantee as that is a resource which is constantly in flux and due to the multitude of factors which are completely out of our control. One moment you may get full speeds, the minute later you are getting 50%. Due to this sometimes quite significant time-dependent variable here, we have to exclude this particular metric from our guarantee.
- Whenever we go through a major generational change in hardware, which will probably happen every 2-3 years or so, you will be offered to be migrated to the new profile which has the equivalent cost as the one you are on now. In short: automatic upgrades for free. If you decline, you will get your "old" hardware profile at a discount.
If you have any questions or concerns regarding our Performance Guarantee, please be in touch with Webdock Support.
How do we do it?
We get this question asked a lot. The industry is so used to optimizing tenant (tenant=hosing customers) density and is so marked by a reseller culture (It's resellers all the way down!) that people don't realize that us "near-first-tier operators" who own our own hardware and are in co-location can take advantage of economies at scale and low overheads. Webdock has perfectly adequate profit margins even with an outstanding hardware mix and high resource allocation to our users. What we do specifically is:
- We operate high-quality refurb hardware (typically Dell Poweredge units) in order to cut costs on outlay while providing excellent performance
- Take advantage of Linux Containers as our hypervisor is LXD. Many providers would see this is a good way to increase tenant density due to the much lower overhead of Linux Containers vs. traditional Virtualization. At Webdock we do the opposite: We pass on more hardware resources to you, the end-user.
- We actively balance and adjust capacity according to load. No host is ever allowed to go above 30% utilization sustained. This balancing happens 8 times every 24 hours, so if you provision a new server you will always be placed on a node with plenty of capacity and which is not under any high load.
- We really do keep a close eye on customers who use a lot of resources as really, in our experience, really high load in about 90%+ of cases is really just bad news. I.e. users doing shady workloads such as spamming, network scanning, cryptomining and other such nonsense.
- For these reasons, if you think your website is really popular and needs a lot of resources and you have maybe been disappointed with performance on other VPS providers, you can trust that Webdock will most likely provide everything you need, and more, for your legitimate workloads.
- Compute density is increasing rapidly these years. Within a few years we are looking at deploying kilothread (1000+) Epyc systems in a 1U package. That's a lot of flippin' compute.
- And frankly, maybe we are just not as greedy as so many other providers ? ¯\_(ツ)_/¯