Rackspace office, London.

Our hosting stack from the bottom up - Part 1

Choosing an infrastructure partner, it's the foundation of everything that follows. It's vital you get it right. That's why we cast our net wide. Read on to find out more.

Photo of Greg Harvey
Wed, 2015-10-07 18:16By greg

Update: We're not in the business of pulling content, but this is an important update on our current private cloud supplier. As of October 2016, Code Enigma has moved private cloud services to Pulsant. This should not be seen as a reflection on Rackspace in any way, just a case of product "fit" and our business needs. Most of the below also applies to Pulsant, if it didn't we would not have moved our services there, so the content of this post is still largely valid.

§

Over the next few weeks I'll be writing a series of blog posts explaining the Code Enigma stack, the technology and the motivations, from the bottom up. We'll be covering things like our virtualisation platform, our choice of Linux distro, our deployment tools, testing suites, etc.

But starting at the start, the base of everything is our hardware and networking. Choosing a partner for all that stuff is a pretty big decision, not to be taken lightly. We've made quite a bit of noise about being Rackspace Gold Partners, and some people might think we just picked Rackspace because, well, it's Rackspace. Nothing could be further from the truth. So how did we end up here?

For us, with the accreditation we carry and the contracts we have in place, "under someone's desk" is not a sensible option for our servers (though you'd be amazed how often that is still a thing). We set about an extensive procurement process to select a supplier for our hosting business. In the end, after talking to dozens of companies (and a few consultants) we invited six different hosting companies to tender, from the less well known (one company was a tiny Dutch firm with a very interesting product and a great technical team) to the big players. We scored them carefully against a range of criteria, both technical and commercial, to select a top two for a "final" run-off, and ultimately a winner. It was a really close run thing between the final two, and the other finalist wasn't a "big name" provider either.

So let's look at some of the technical "whys" for going with Rackspace.

Firstly, there's the datacenters, all designed to minimum Tier 3 principles (Rackspace ignore actual Tier 3 certification and prefer to release a statement of design principle adherence instead) and their new LON5 facility in Crawley - the home of our private cloud - is probably the most state-of-the-art facility in the UK. And of course, that's just in the UK. There are facilities all over the globe, including Australia, Hong Kong, several in the USA, and so on. All of which we can make use of, should the need arise.

It's also important to us the datacenters are ISO 27001 certified, though admittedly, this isn't a differentiator for hosting companies these days. It's hard to find a company running its own datacenters that doesn't have ISO 27001 certification - but it's always worth checking! There's no point in us certifying our systems, then placing them in an uncertified facility where the security standards are unknown. (Just as there's no point in using an ISO 27001 certified datacenter and putting non-certified products in there!)

Then there's the networking. (And this is one of Rackspace's biggest wins - one we've already experienced.) Rackspace have no less than nine backbone providers - I've seen their EU backbone control centre and it's impressive stuff. If their network is down, the Internet has far bigger problems! Indeed, we've already had an incident where we were seeing packet filtering on a connection upsetting an IPSEC VPN and Rackspace were able to simply shift us to another route and avoid a problem ISP entirely. I don't think anyone else we spoke to had the kind of reach and influence Rackspace has when it comes to working with peer companies. That's why they have a 100% network uptime guarantee.

Then there's the Rackspace product catalogue, which is simply huge. We can do anything from straight public cloud infrastructure (file storage, load balancers, server instances, think AWS), to dedicated bare metal, or anything in between. There is practically nothing we cannot do.

If a customer comes to us with their own search appliance they want to co-lo (Rackspace have a very simple policy - if we don't sell it, you can co-lo it), a requirement for bare metal database servers, app servers in a VMware private cloud cluster with highly available SNAS storage, and a layer of auto-scaling public cloud reverse proxy cache servers, plus a dedicated fibre connection from our cabinet to their office comms room for management, we can do it. We can actually do it!

I still have to pinch myself, having migrated from a major public cloud provider. We're not used to having such an amazing toy box at our fingertips. Sure, with AWS for example you have a load of excellent cloud services, but that's it. That's where it stops. With Rackspace you have a similar suite of products and a ton of other custom and specialist products, plus the ability to host your own in Rackspace premises if they can't sell it to you.

"How often do you need all that?!" I hear you cry. Well, not often, to be fair, though we have a few customers who do have similar requirements.

But the wider point is who knows where our business is going in the future? And who knows where our customers' businesses are going in the future? By way of an example, we just had a huge year one product review where I went over to the Rackspace office in London and sat with our team there to discuss our platform options. We stuck with VMware (more on that in the next installment) but there were four virtualisation platform options on the table for our consideration. Now that is a product catalogue for future-proofing your businessWhen it comes to planning your next move, you want to be working with the people who have all the avenues open, right? Rackspace will find a way to support us, no matter what we want to do, and that's priceless for us and our customers.

Aside from the technicals, there were also some "soft" wins we felt Rackspace had over the competition, namely brand value and ownership.

Brand value, that's an obvious one. Love them or hate them, everyone in tech knows Rackspace. And the vast majority of the people I know, either through direct experience or through reputation, see them as a safe pair of hands. So yes, we felt the value of the Rackspace brand, being backed by those people who are the quintesential "safe pair of hands", was compelling.

However, it goes deeper than that. Initially (and perhaps somewhat cynically) we thought the name Rackspace might be a box ticked for some potential customers. We rolled our eyes at "Fanatical Support" and the way everyone is a "Racker" - it still is some cause for amusement on occasion. But now we've been Rackspace Gold Partners for over a year, we can confidently say to our customers Rackspace have our back all the way. Backed by Rackspace we really can offer people the best of everything.

At the sharp end you have Code Enigma, we know Drupal, we know Linux, we know the software stack to keep your applications secure and performant, and we know how to manage it efficiently and securely. And behind us you have Rackspace, who know networking, they know hardware, they know physical security, they have real influence with global infrastructure providers, they can reach into the guts of the Internet to fix stuff for us if they have to (and they have in the past).

That's not just marketing, it's not just a procurement "heard of them!" checkbox - our real Rackspace experience gives us real confidence that, with them, we can deliver. Confidence that we're absolutely within our rights to tell would-be hosting customers we are an A-Team, Rackspace and Code Enigma. We will get the job done, and we'll do it better, quicker and more professionally than almost anyone else in our sector. But Code Enigma couldn't do that alone, we have that confidence because we have Rackspace, and on balance Rackspace have earned it.

It's a big machine, sure - but another one of the interesting things that made us pick Rackspace was although it's a big machine, they were able to convince us we wouldn't have a "big machine" experience, and that has born out. We don't feel lost in the Rackspace customer base, stuck in a forgotten corner with an account manager who did something wrong. Far from it, we feel integrated (we're engaged, we meet their marketing people, we get invited to their events) and valued (we have a dedicated tech team, a dedicated account manager, we're on first name terms with engineers) and we get really good service. And that is not easy to deliver consistently when you're the size of Rackspace! (Also, I might add, huge hat-tip to Sab Knight and James Waggott who did so much to make our first year at Rackspace a good experience.)

And ownership? Well, unlike some competitors, Rackspace is still (at time of writing) an independent organisation, yet to be swallowed up by venture capitalists or large competitors (although they did review their options in spring of 2014). This is not so important, but having said that, it does provide a degree of stability others may lack. Many smaller hosting companies are lurching from VC to VC, have frequent changes of management and, with those, come inevitable changes of strategy, direction and other senior staff. Senior staff you may have come to enjoy a solid professional relationship with. At Rackspace, everyone we've worked with has been there for years! And for all the right reasons. (They are consistently voted one of the best places you can work.)

So there you have it. Layer one of the Code Enigma hosting product, the hardware and network support we get from Rackspace. Watch this space for the next piece, our virtualisation platform and why we chose it.