Streamline Computing has been building and integrating clusters since 2000. Our philosophy has always been to make quality, reliability and supportability the hallmarks of a Streamline cluster.

Over the years this has driven a number of 'best practices' in the business; we select hardware and qualify it for cluster use based on its reliability and suitability whether it is a server, switch or high performance interconnect. Part of the qualifying process includes testing the component against the installation and performance of Streamline's HPC software stack and ensuring that not only does it pass the performance test but that it is also supportable for the life of a cluster (typically 3 years). On occasion this means that Streamline may prefer not to supply particular models and variants but as part of our value to our customers we can say why not and recommend 'best-of-breed' solutions based on over a decade of experience and the delivery of over 500 HPC systems.

Highly experienced software and hardware engineers at Streamline are responsible for the development and maintenance of Streamline's HPC software stack. It is a product, assembled from the highly fragmented Linux open source and proprietary landscape and currently based on RedHat Enterprise Linux, that is constantly being updated, tested and installed on nearly every system Streamline ships. Streamline's software stack represents over 50 man years of development, testing and refinement and it ensures that every Streamline system it is installed on, still satisfies our goals of quality, reliability and supportability.

---

Our facilities

Based in the Warwick Technology Park, Streamline's facilities are one of the things that makes a real difference to our customers. We have built a solid track record of being able to deliver systems quickly and efficiently; typically even the most complex of systems are up and running within hours of being delivered to site by our engineers. So how do we manage this when the marketplace is littered with systems that have taken months to complete?

Back in 2006 we invested in an Integration Centre; this is 270 square metres of factory space equipped with sufficient power and evaporative cooling to house some of the largest HPC machines built in the UK.

Building systems in house allows us to test and configure all systems in a controlled environment with access to all of our engineers and equipment before delivery, reducing time on site to commission and eliminating early hardware failures once a system has been delivered.

Most cluster components that fail do so within a few hours of the start of their operational lives so Streamline strives to cause this to happen before delivery and while still the system is still in our Integration Centre so that replacement can be performed rapidly and the component retested. This way, systems shipped by Streamline are truly run in and go into production at the customer's sites very quickly. This is only possible because of the facilities Streamline has invested in to build and test systems prior to delivery.

---

Our customers often visit to check on progress, run tests and undergo final training before the system is broken down and packed before being shipped to site for re-assembly. It required a lot of investment (which is probably why we are in a minority with this facility) but it has definitely been worth it .