This page contains a Flash digital edition of a book.
HPC 2013-14 | Te cloud

Kress, CEO at Coraid, believes that as clouds mature and become more widely deployed and used, the industry will see changes in demand and requirements. ‘Customers want to be able to seamlessly move workloads between private and public clouds depending on level of use, resource constraints and the need for tight security and control. Terefore, the demand for tiers of cloud services at different levels of price-performance and resiliency will go up,’ he said. Kress believes people will look for

policy-based automation and orchestration tools to help them deploy workloads on different classes of services and as they start using clouds for mission-critical applications, we will see more requirements around enhanced security, fault tolerance and control. Tis will hopefully alleviate some lingering concerns regarding security. Te nature of how users interact with

the cloud, and indeed its function, is also expected to evolve, and DDN’s Jean-Luc Chatelain predicts that mobile devices will have a significant impact. ‘Tis is not to say that people will be running algorithms on

smart phones, but many of the tasks that are connected to the cloud will happen locally.’ He explained that if a device is connected then a large proportion of the compute will occur on the cloud, and when a device is not connected, some level of compute can still be done locally until the connection is re-established. ‘Tis is driving the evolution of the cloud

into being a massive data repository,’ he added. ‘I foresee a time when all referential information will be held in the cloud and that all pre-processing will be done locally before being pushed out. Tis is why mobile devices will become an integral part of the process.’ Andrew Carr, CEO at Bull UK and

Ireland, suggests that in the next three to five years, three distinct streams will be merged and consumed as a service: traditional HPC, data management and storage, and high performance analytics. ‘It will ultimately have to become a commoditised market and we are currently at the very embryonic stage of the process,’ commented Carr. ‘One challenge for taking HPC to the cloud or on-demand model

is the education of customers to want to buy a service rather than publish a bill of materials or kit list, which makes it a price negotiation rather than a question of value. Tere’s a significant amount of general awareness that needs to happen for this type of model to be successful.’ Given current adoption rates of cloud

services, companies like Cycle Computing believe that over the next few years, technology and applications changes will drive users to the cloud. According to CEO Jason Stowe, capability supercomputers and fast interconnects make a lot of sense for some applications and HPC in the cloud will never replace that. However, there is a new form of capability machine emerging that has radically different properties, such as scalability, metered usage, and access to workloads that are cost prohibitive to deploy on internal HPC clusters, and that these aspects will all increase the demand for cloud services. ‘Ultimately,’ offered Stowe, ‘most research

organisations can pay $4,400 but not $40 million, and that alone will ensure cloud continues to have a place in HPC.’ l


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36