Some time ago, I wrote about how DataLords works.
In just 8 months, we've had significant changes, so I thought it was worth writing about here.
In March 2022 we gained investment and new owners of the company. This allowed us to spend where we really needed to spend, and invest in both software and redevelopment of our own platform. We also invested in new networking hardware, meaning our core network doubled in speed and is now powered by Cisco.
So... how does the DataLords platform work now, and how can our platform benefit your business?
One of the key things I was never happy with, was that we had High Availability configured on our previous platform, but that did mean in the event of a hardware failure, a percentage of websites would be down for 2-3 minutes. For most hosting companies this is perfectly acceptable, and their SLA's actually allow for much longer downtime over the period of a month.
But we're not most hosting companies.
Originally, we were using software called Proxmox. To try and avoid explaining every little detail - to summarise - Proxmox allows us to use our physical servers to run multiple virtual servers. This software was great but as I said, in the event of a physical server failure - there would be downtime.
So we decided initially to completely move from Proxmox to VMWare. VMWare allowed us to rethink our deployment strategies completely, and this resulted in a complete overhaul/rewrite of our back end management platform; in the end, we've deployed a hybrid solution - utilising some Proxmox partly as a datastore but partly for running some services on.
i.e. In an ideal world we'd run everything through VMWare, but the costs would end up making this non-viable for many businesses, and whilst we won't be right for everybody - we certainly wanted to make sure we maintained our ethos of 'dedicated hosting for the price of shared hosting'.
Ultra High Availability
So we introduced a new product category. Ultra High Availability (UHA) which runs inside our VMWare platform. The likelihood is you haven't heard of Proxmox, but you have heard of VMWare. They've been around forever, and have specialised for years in providing enterprise grade services around virtualisation and much much more.
VMWare introduces us to a new feature: Fault Tolerance. This works by essentially running the virtual machine on two physical servers and maintaining consistency between them. In the event of hardware failure, the platform boots up another copy of the virtual machine on another server, and resilience is restored. This all happens automatically and there is absolutely zero downtime to the service.
The consequence of this is, of course, that it uses literally double the resources - hence the cost premium.
So we balanced that with our High Availability platform.
High Availability
Because of the move to VMWare, we restructured everything in our platform and changed how we worked completely. Originally with Proxmox we used their API to create Linux Containers (LXC) and managed the resource allocations that way. After we changed things around with VMWare, to enable the Fault Tolerance system to work, we ran multiple Virtual Machines and configured them as LXD Hosts (the LXD Hosts run the Linux Containers (LXC)).
This actually gave us a lot more flexibility and control as we weren't subject to the Proxmox API features, and could work with LXD/LXC directly. This made things a lot easier to manage, but also gave us the other benefit... Let's say we have 50 LXC environments running - well, on Proxmox we'd have to move them one by one, and whilst they wouldn't really have "downtime" they would have to be "rebooted". This meant that sessions were terminated/interrupted.
Not ideal for a busy e-commerce environment.
Although for most hosting companies - this would be acceptable they would just (typically) move machines outside of business hours or overnight - this wasn't what we wanted.
But by enclosing these LXC environments into a virtual machine, we can "live migrate" the whole virtual machine from one physical server to another and have no interruption at all. This allows us to do maintenance at any hour of the day without any concern for downtime or interruptions.
So, the upshot of this is...
The real benefit is - within each package we offer, we have a percentage of sites that are UHA sites.
So if you're a busy e-commerce site, or you're an agency and have VIP / busy sites - you can put them on the UHA platform. Regular sites can sit on our HA platform (which is still extremely robust and we offer 99.99% SLA) and you get all this for - generally speaking - the equivalent you'd have to spend to host the same number of sites on dedicated machines with something like cPanel.
Talking of cPanel, we originally used WHM/cPanel for managing Email and DNS as this was an "easy" solution to get started, but with our investment and plans to scale up - we wanted something that we had a lot more control over, and something that would scale using fewer resources.
So we've built a completely custom mail platform, with dedicated hardware to run virtual machines for back end mail stores, front end mail servers, MX servers and webmail servers. This alongside our bespoke DNS platform means that we're ready to take on 10 sites, or 10,000.
If you're interested in finding out more about how we can help your business, call us on 020 7870 8025, or join the waiting list here.
