Origo Compute – IaaS for enterprises

Jun 6,2015 No comments yet By Origo

We will soon be launching our new compute service into a market which is dominated by huge players with near infinite ressouces, specifically Amazon and Google. We do however not think their compute products are very well suited for the typical enterprise client.

In this blog post we try to explain our views a little. You may not agree – traditional enterprise computing is of course not all bad, and there may in fact be enterprises running their mail servers on Google Compute. If so, feel free to drop a comment..

Consumer vs. enterprise computing

Consumer computing has developed explosively over the past decade, with the iPhone, Android and iPads and the consumer computing backend infrastructure, collectively known as “the cloud”. We think it is safe to say, that very few people miss the old ways. Downloading and installing software, syncing via a cable connected to a PC, downloading and installing upgrades to Office, setting up and running scheduled virus scans, the list of things that look obsolete for the average 17 year old iPad user browsing around the App Store goes on and on. A large part of this effective new computer paradigm is enabled by highly effective back-end computing, driven by large custom-made server farms in efficient data centers, with customized software which orchestrates the tasks of scaling, load-balancing, software deployment, fail-over, etc. The new age of consumer computing has given us the most efficient large-scale computing the world has seen so far, with Google, Facebook and Amazon as the poster childs.

Compare this to enterprise computing, which by some accounts has mostly not seen any of the efficiency improvements delivered by consumer computing. Enterprise computing is to a very large extent stuck in the same old server rooms with Windows boxes asking for the daily fix of system administrator quality time full of Windows updates, Active Directory issues, enterprise software patches, etc. On the hardware side, the servers have not seen any real development either over the last decade. The server rooms are still stuffed with Intel boxes, Fibre Channel SAN’s and dedicated switches from vendors like HP, EMC and Cisco, running Windows on VMware. All this has been going on for many years, and there is a large pool of skilled workes who can maintain this kind of infrastructure. The efficiency of cloud computing has however made it clear, that things will and need to change. It is somewhat telling, that of all the break-through infrastructure delivered by the new consumer computing players such as Google, Facebook and Amazon, virtually nothing is built by the old guard of enterprise vendors. The cloud does not run on Windows, EMC SANs, VMware hypervisors and Cisco switches. The cloud runs on Linux, custom built servers, cheap disks and KVM and Xen hypervisors.

How did it come to this? Enterprise computing was supposed to be all about performance, scalability, efficiency and security. At least if you read the brochures. Well, judging by the current state of things, the brochures were mostly wrong.

Our background

Through the last decade we have made a good and stable living by delivering traditional enterprise computing to clients in Denmark, where we are headquartered. Our team of system administrators have done and still do most of the above on a daily basis. This will not change overnight. But – a couple of years ago, growing increasingly frustrated by things like increasing license complexity by vendors and the increasing failure of traditional vendors to deliver any reasonable value for the money they were charging, we began developing our own infrastructure stack. We don’t think this is the end-all answer to infrastructure, it is largely born out of frustration with existing solutions, but we use it, and have run our business on it for over a year now. While much will be improved, rewritten, refactored etc., it is our simple experience, that we are able to deliver much more value to our clients with our “home-made” stack, than with the traditional enterprise stack which we used to use. Our own stack consists of custom server hardware, and an infrastructure stack, orchestrating KVM hypervisors and ZFS storage. It is by no means perfect, but neither was what we used to have – in fact it was often closer to a nightmare. With our own stack we are able to deliver the service we would like, and effectively bring some of the progress of the cloud into enterprise infrastructure. We are simply more profitable this way, and it is much more fun.

Our products

Enterprises are not rushing to place their data in the cloud and running their servers on AWS or Google Compute – there are simply too many very sound reasons not to (including cost, bandwith/latency and security – yes, industrial espionage still exists). The situation is however not stable, simply because traditional enterprise computing is fast becoming so inefficient compared to the cloud, that there is a mounting pressure from management to deliver more efficiency and new services. Without much help coming from the traditional vendors, who are on the contrary increasingly focusing on squeezing every last license dollar out of their clients, IT departments are generally in a tight spot. We hope to help showing some way out of this mess.

In order to run a typical IT infrastructure more efficiently we present two products:

Compute“, which is basically an alternative to having your own hardware and server room, and “Engine“, which are the custom-made hardware boxes we run our compute service on. Buying one or more Engines basically allows you to retain some virtual servers and storage on-premises, while being able to use the exact user interface to manage your infrastructure in the cloud and in your server room.

Stay tuned for more to come.

Leave a Comment

Your email address will not be published. Required fields are marked *

*






Services

Products

Company

Help