Rackspace’s OnMetal servers sound like a contradiction in terms. They are “cloud” servers, but they aren’t virtualised. They offer customer their own bare-metal, dedicated hardware.
That hardware consists of servers built specially for Rackspace by Taiwanese manufacturer Quanta, using specifications from the Open Compute Project – which Rackspace co-founded with Facebook.
The servers are designed for specific tasks, and use an open specification. That in itself is important. But bare-metal cloud? What
It’s been a mantra for a long time that the public cloud is based on virtualisation. By virtualising the hardware, cloud providers can set up new customers quickly and support multiple tenants efficiently, spinning up virtual servers as required and aggregating loads onto non-specific hardware.
But mid-to-large cloud customers have started complaining lately, about “noisy neighbours”. If you share a machine with another virtualised tenant, it can hog the resources when it launches a complex process.
Providers have also started to realise that virtualisation adds a layer of software – the hypervisor – which consumes resources of its own.
But what would a bare metal cloud server look like? What would it run? The answer, it seems, is coming from OpenStack.
Ironic is a bare metal provisioning system. It’s not actually ironic at all. It aims to let service providers dish out bare metal servers with great flexibility, and Rackspace is bringing it into actual use.
So if you are a Rackspace customer, what do you put on your bare-metal server? If you then virtualise it and clog it up with your own virtual machines, what have you gained?
The answer may well be to use containers. They share resources at the application level, putting multiple apps on the same host OS, each in a separate fenced-off ‘container’, not a whole virtual machine.
The lead standard here is Docker, and it’s got support from Red Hat, Google and others. A new Linux version for containers called CoreOs has emerged, and is among the options Rackspace is offering.
So we have flexibility on a lot of levels. Open hardware, with the Group Hug standard allowing a choice of processors, and Ironic allowing them to be deployed more flexibly. Docker allowing containers instead of whole virtual machines and CoreOS supporting that.
Rackspace is trying to find a niche to distinguish its services from the cloud giants like Amazon Web Services. If it can’t do that, it might as well follow the path it engaged Morgan Stanley to investigate – of merger with larger players.
OnMetal is a new kind of cloud service, and it could be a very interesting move – for Rackspace and its customers.
Recharge your renewable energy knowledge with our quiz!
A version of this article appeared in Green Data Center News
Suspended prison sentence for Craig Wright for “flagrant breach” of court order, after his false…
Cash-strapped south American country agrees to sell or discontinue its national Bitcoin wallet after signing…
Google's change will allow advertisers to track customers' digital “fingerprints”, but UK data protection watchdog…
Welcome to Silicon In Focus Podcast: Tech in 2025! Join Steven Webb, UK Chief Technology…
European Commission publishes preliminary instructions to Apple on how to open up iOS to rivals,…
San Francisco jury finds Nima Momeni guilty of second-degree murder of Cash App founder Bob…
View Comments
Adherents of the virtualized cloud would like us to believe that they offer the only game in town for enterprise scalability, but that’s not the case. With the exception of a few edge cases where by-the-hour elasticity is required, bare metal clouds provide levels of scalability more than sufficient to meet the needs of most businesses, while also providing more efficient use of hardware resources, better reliability and availability, and superior performance.
The company I work for, InterWorx, our control panel provides an all-in-one bare metal cloud platform for managing web servers, database servers, and file servers, with sophisticated cluster management capabilities, support for multiple remote databases, advanced load balancing, and realtime resource graphs of cluster nodes.