27 Feb 2012
Five short stories from the cloud show how platforms are maturing and illustrate that despite all the talk about virtualisation and mobility there is still good old-fashioned hard-wired physical infrastructure behind it all.
GoGrid – old hand from the valley now in Europe
GoGrid has recently announced its first European cloud infrastructure services, provided from an Equinix colocation facility in Amsterdam. Its name might not be that well known in Europe, but GoGrid comes with pedigree. It was founded in 2000 in San Francisco and claims to have become the “number one dedicated hosted service provider” to the Silicon Valley elite. Through the experience learned over 12 years as a managed hosting provider, it has built its own “Cloud Infrastructure Stack", an infrastructure-as-a-service (IaaS) platform, which enables the delivery of hybrid hosting services; clouds that consist of discrete physical infrastructure and public cloud resources all managed through a “single pane of glass” interface.
But, why bother with your own infrastructure?
CloudBees – taking platform-as-as-service (PaaS) to a new level
Getting someone to build your organisation a private cloud from scratch is one approach. But, how about turning existing datacentre infrastructure into a private cloud and making it easy to extend by adding on-demand resources from established IaaS providers. CloudBees' AnyCloud aims to do just that. It was founded a little under two years ago by a team of IT industry veterans including CEO Sacha Labourey (ex Red Hat/JBoss). AnyCloud is a Java-based PaaS offering that is layered over existing infrastructure; either that already owned by a given organisation or other IaaS offerings such as Amazon EC2, Rackspace Cloud Servers or any other local IaaS provider. Once set up CloudBees undertakes to manage it all for you.
And if you are UK based that local IaaS provider could be Attenda….
Attenda – local UK provider ups the ante
Attenda’s IaaS offering, known as Attenda RTI, is sold alongside dedicated managed hosting services all based in the UK. On top of this has been added a business-focused professional services overlay. Attenda has observed (as has Quocirca), that line-of-business managers are increasingly involved in the decision to purchase cloud-based services. This is particularly true in the mid-market where Attenda is focused. Mid-market managers know they need applications, but are not so sure they need to worry about the infrastructure to run them on. So, Attenda has launched an initiative it is calling "Business Critical IT” that combines a structured business engagement methodology with recommendations for supporting infrastructure and services. Attenda says this addresses the need to focus on business outcomes rather than technology ones; Quocirca would not argue with that as an objective.
But ultimately someone has to run infrastructure…..
City Lifeline – baked into the heart of London
The big co-location and managed hosting providers are always keen to show off their state-of-the-art, usually purpose-built datacentres on sprawling out-of-town trading estates, for example in Slough or London Docklands. But, just as impressive is to see how, in order to deliver the low latency and physical proximity required by financial services organisations in the City of London, City Lifeline has squeezed in 28,000 square feet of datacentre space in Hackney, just a stone’s throw away from its key customers. This is no purpose-built facility but an older building that has been adapted; finding and paying for appropriate space in such a central location would be prohibitive. Proximity allows City Lifeline to charge about a 30 per cent premium over that of out-of-town providers. However, despite these seeming limitations it is still expanding on the current site by building over its small back yard. It is not just the difficulty of finding suitable locations in the City that keep it where it is. City Lifeline is hard wired into the heart of London; the datacentre sits right on the backbones of 22 internet and voice carriers, for all of which City Lifeline hosts points of presence.
Actual cables may never go away, but you can reduce the number…
Xsigo Systems eliminates miles of cables
Observe the rows of cabinets in most datacentres and you will see thousands of Ethernet cables linking the individual rack units to each other and to top of rack switches and each cabinet to end of row switches. All these cables are linking servers with the infrastructure they rely on; storage, load balancers, network routers, security appliances and so on. Now it is possible to eliminate many of these cables with the Xsigo Server Fabric. It uses up to 40GB InfiniBand cables to connect each server and each peripheral device to a Xsigo fabric appliance, which takes up two or four RUs and acts as a broker between all the various bits of hardware. Furthermore this means that once implemented, the Xsigo appliances see and collect all datacentre traffic and can act as a feed to performance monitoring tools. This has led to the vendor’s latest announcement of the Xsigo Performance Manager. Eliminating so many cables saves money and space, but also increases performance as its customers seem happy to testify.
Bob Tarzey, Analyst and Director, Quocirca Ltd.