Are you wondering which infrastructure would be most suitable for your line of research? There is no general solution. Our consultants always start by making a recommendation tailored to your specific infrastructure needs. Below are a number of considerations that play a role in making a tailored recommendation.
1. Large volumes of data or extensive computational tasks?
Do you need to analyse large volumes of data or are you more interested in powerful computational applications, such as those required for simulations or modelling?
Are you dealing with large numbers of computational tasks that can no longer be carried out at a local or centralised cluster in a short time frame?
If so, you may be interested in our high-throughput data processing services. All infrastructures have their own dedicated storage systems. Our Visualisation Service is available for researchers wishing to visualise their results.
2. Capability or capacity computing?
When choosing between various computational infrastructures, it is important to understand the difference between capacity computing and capability computing. Lisa Compute Cluster and Grid provide capacity computing, while the Dutch National Supercomputer Snellius delivers capability computing. But the difference between the two lies not so much in the processors (which are around the same speed). A capacity system, as used in Lisa Compute Cluster, enables you to perform extremely large calculations, but with 1 fundamental limitation: the calculations, models and algorithms must be segregated and run independently from one another.
With capability computing, we combine the most powerful processors, a large memory capacity, high inter-processor communication speeds and ample and fast disk storage space. Capability computing is a particularly essential tool for simulations and modelling where the speed of the connections is just as important as the speed of the processors. Where this type of computing power is required, the Dutch National Supercomputer is a logical choice.
3. Ease of use or flexibility?
The infrastructures also differ in terms of ease of use and complexity when performing computational tasks. The Grid requires special tools to distribute compute jobs to the various clusters across multiple locations. This is precisely what makes it the most scalable infrastructure available with almost unlimited capacity. The Grid, however, requires more technical know-how on the part of users in particular, compared with other infrastructures. HPC Cloud places responsibility for the management of your virtual machines in your own hands. This service is relatively accessible to users with a moderate level of expertise.
Another consideration is the availability of software. The Dutch National Supercomputer and Lisa Compute Cluster feature a lot of pre-installed software, such as the standard software used by chemists and fluid dynamics experts for their analyses and simulations. We can also install software for you on request. HPC Cloud gives you more freedom in installing and removing software, which brings us to our final consideration: flexibility. HPC Cloud offers you the greatest degree of freedom and flexibility. Not only are you free to choose which software to use, but also the operating system. Linux is the designated operating system for all other infrastructures. SURF will of course endeavour to meet your requirements no matter what system you choose.
Advice and consultancy
Our compute infrastructure specialists would be happy to help you select the solution that best suits your research project. Feel free to contact us at firstname.lastname@example.org.
In need of independent advice on the optimum usage of SURF's computing and storage facilities? Our consultancy service will be happy to help.