Wable - New VPS Review

Wable is a new VPS provider with a twist. Instead of buying each individual VPS, you buy a "bundle" of resources that you can distribute amongst up to as many VPS as your bundle allows.

Why Bundle's Are Better

Imagine you have a website that converts videos. This website may need three services:

  • A web frontend that just displays information to user's and allows them to upload videos, as well as tracking all the information in the database of your choice.
  • A storage service that is just responsible for storing and retrieving the videos.
  • 1+ "compute" engines that actually performs the conversions by pulling one video at a time from the storage service before converting it and sending it back.

Each of the services listed above are completely different in their resource requirements. The storage service mostly needs storage, the database needs lots of RAM and some CPU, and the compute needs all the CPU that the database isn't using. With the bundle this is easy, you can just deploy your 3+ services on their own VPS and allocate the resources as they need. On most other providers, you would have to buy one of the larger plans for each VPS in order to get enough of the resource you actually need, which costs you far more.


Wable only uses local RAID 10 SSD storage which is the same as Digital Ocean. However the price per GB is cheaper. There is a limit of 100GB storage per VPS though, which you would have to workaround with something like GlusterFS if you need a much larger volume. Providers like AWS and Rackspace only use networked storage, except for their ephemeral drives which almost makes them pointless.

Digital Ocean Price Comparison

Below is a comparison of two plans that are as close in price as possible with Digital Ocean on the left and Wable on the right

As you can see, paying 20% more (just $1) gets you:

  • 2 x CPU access
  • 3 x bandwidth monthly utilization limit
  • 2 x the storage capacity
  • RAID 10
  • 5 additional IP addresses

The other factor to think about is the access to a 2-10 Gbps port. I need to perform further testing, but I'm pretty sure on D.O. that I was capped at 100mbit.


Before rushing off to migrate all your infrastructure, there are a few things that you should think about. These are OpenVZ based VPS's rather than KVM (Digital Ocean) or Xen (Amazon Web Services/Rackspace). This may not be an issue for you, but you should run a trial bundle first to see if there are any kernel modules you need which aren't provided in the 2.6.32-042stab085.20 kernel that they are using at the time of this writing. The other big factors are that, being new, they have less of a reputation to judge them by, and I do not see any details relating to how much compute power you should actually expect to have. E.g. client's are limited by core access only, and there is no way of knowing how "oversold" each node is.



The most important part of any product is the price, and as described earlier, their price-to-resource ratio appears to blow away the competition.


Wable's bundle system whereby you buy "resources" and allocate them between multiple VPS's is completely unique. This is great if you have a variety of services that have different needs. E.g. your backup service just needs lots of storage, whereas your database probably needs much more in the way of RAM and CPU.

Other providers do allow you to scale a VPS's resources individually, but at exorbitant prices. What is unique about Wable's concept, is that you buy a bundle, and then split that as you like across your servers. This probably helps keep their price's competative.

5 Second Scaling - No Reboot

It turns out that you can change the resource allocation setup of your VPS whilst it's runing (no reboot), and the process will take just 5 seconds as demonstrated in the video below:

This is incredibly useful. You can now pull resources from your other servers in order to quickly respond to demand, e.g. such as scaling up the CPU on your webserver/database when there is a burst of traffic. As soon as there is an API, you can have code monitor this for you adjust automatically. The key factor here being that there is no downtime. This cannot be done on Digital Ocean, Amazon Web Services, or Rackspace. They all require a reboot, and some resources on some services cannot be scaled.


This is actually two points combined. Using only SSD storage on its own is a huge factor, which has been one of the selling points of Digital Ocean. With AWS/Rackspace, which use networked storage, you may notice a significant percentage of your CPU time being wasted on "disk wait". Nothing compares to using local SSD storage in terms of latency. IOPS and throughput can vary a great deal on how oversold the node is, but are also generally better. Being RAID 10 means it actually uses twice as much in physical disk storage to bring you that capacity in order to give your data redundancy protection whilst also being split across drives should hopefully help with disk throughput. It doesn't state on the site if they are using software raid, fake raid, or true hardware raid (LSI/3ware).

Disk throughput depends on so many factors, including block size, filesystem type, node contention, and the bandwidth limits on each intermediary physical layer (SATA expanders/RAID cards etc), that all you can do is benchmark, which I will do later.

Stupifying Network Speeds

Please refer to the network benchmark, but on a very simple test, I did actually achieve a 700mbit download speed. The trick is to find an external server that can match the same bandwidth levels. There is no point performing tests from within the Wable network.


OpenVZ based

Using OpenVZ basically means you share the same kernel as the host and everybody else. As some people put it, this is very much like a glorified chroot. This is unlike Xen and KVM where you can use any kernel you like. This may not seem important but if the host does not have a kernel module that you need, then bad luck. This particularly affects me because everything I create now runs on docker, and with docker, having the latest kernel really helps, especially as it is delving into areas such as BTRFS based storage engines etc.

You may also have a tough time setting up the firewall as you like, and have to go back to good-old-fashioned IP tables.

OpenVZ is perfectly fine for Minecraft servers.
Wable will not currently run docker containers. OpenVZ is generally not suitable for running docker, although I'm sure that there are a few specialists out there who managed a workaround, but this requires you to be able to update the kernel, something you cannot do as the client. If you need to run cheap docker containers, Digital Ocean is your best bet.

No Attachable Networked Storage

Unlike Amazon Web services and Rackspace, there is no attachable networked storage. However, the flexibility of allocating resources as you like from your bundle, slightly offsets this factor compared to the likes of Digital Ocean's setup. E.g. buying a larger bundle doesn't feel like a waste as you can allocate the extra CPU/RAM to your other VPS's

No API - Completely Manual

This is rather surprising since it seems that every cloud service provider has one. An API enables developers to deploy/scale their infrastructure quickly/automatically. Ideally your services should automatically scale with demand, rather than wait wait until the admin gets an alert.

America Only

This service is provided only in America, which for some of us, is the last place we want our data/services located due for legal/latency reasons. With Amazon Web Services, you can have your services deployed pretty much anywhere in the world, and Digital Ocean have datacenters in New York, San Francisco, Amsterdam (Europe), and Singapore.

How Much Compute Power Will I Actually Get?

Wable is very much in the same boat as Digital Ocean in terms of compute. They grant you access to more vCPUs but do not guarantee you any compute power. You just have to hope that the node isn't heavily oversold, and that you don't have "noisy neighbours" sucking up the processing power. This is one of the strengths of AWS EC2 as the ECU unit is a measure of your dedicated compute power, but you pay far more and actually end up with a poorer service. With a $15 budget on EC2 you know you will get a crap service, but you know exactly how crap, because it's guaranteed to be that crap.

8 VCPU per node

You cannot have more than 8 cores on an instance with Wable. This means that if your applicaton is not horizontally scaleable, you may have to go elsewhere. Digital Ocean offers an instance with access to 20 cpu cores, whereas AWS offers a 32 vCPU instance (more importantly 108 dedicated "ECUs" of power). If you need need more than 8 cores, you're probably better off with a dedicated server.


Before looking at my benchmarks, it may be worth looking at the Serverbear benchmark I ran on the budle1 option.


I don't consider this a comprehensive bandwidth test, as I think I may have actually been limited by the sender, or perhaps the disk throughput rate, but I just managed to download the entire Ubuntu 12 Desktop ISO in 14 seconds, which averaged out at roughly 60Megabytes per second. At one point I saw nload reach over 700 mbit/s.


When creating a 10 GiB virtual block device, with a block size of 1MB, I was able to achieve 111 MB/s.

However, I was rather surprised to find that I achieved nearly double this on my $5 Digital Ocean VPN:


Based on these "too good to be true" prices, this could be a very welcome addition to the VPS market. However, time will tell if the service is economically viable enough to survive and remain stable, or just push other competitors to reduce their prices. It takes time to build up a brand reputation/recognition which brings in the business-level consumers who have the real money. This price point looks to be bait for early adopters until they can reach that point. Not having an API is a major drawback, but one that I believe can be quickly resolved.


  1. Terrible service. Avoid them at all costs. They advertise un-metered bandwidth, but if you happen to use more than they like, they will ban your account and say it's because of a "DDOS". They are the most unprofessional company I have ever worked with. I think it's more of a one person show, where the owner acts like a bratty teenager.

    1. It's a shame that you had such a terrible time. I haven't used the service much since trying them out when I heard about them. DDOS attacks are becoming more of an issue and I guess a service that advertises unlimited bandwidth is going to have a hard time cracking down on the activity accurately.