...

We are concerned here with the gathering and scrutiny of requirements for optimisation. More pragmatically, we are concerned with how we might develop generically applicable methods by which to optimise the research output of environmental science research infrastructures, based on the needs and ambitions of the infrastructures surveyed in the early part of the ENVRI+ project.

Perhaps moreso more so than the other topics, optimisation requirements are driven by the specific requirements of those other topics, particularly processing, since the intention is to address specific technical challenges in need of refined solutions, albeit implemented in a way that can be generalised to more than one infrastructure. For each part of an infrastructure in need for improvement, we must consider:

...

Many optimisation problems, whether explicitly identified as such by RIs for , or implicit in the requirements of for other topics, can be reduced down to ones of data placement, often in relation to specific services, resources or actors.

• Is the data

...

• needed by researchers available

...

• from a location

...

• such that they can be easily identified, retrieved and analysed, in whole or in part?
• Is it feasible to perform analysis on

...

• data without substantial additional preparation, and if not, what is the overhead in time and effort required to prepare the data for processing?

This latter question in particular relates to the notion of data staging, whereby data is placed and prepared for processing on some computational service (whether that is provided on a researcher's desktop, within an HPC cluster or on a web server), which in turn concerns the further question of whether data should be brought to where they can be best computed, or instead computing tasks be brought to where the data currently reside. Given the large size of many RI's primary datasets, bringing computation to data is appealing, but the complexity of various analyses also often requires supercomputing-level resources, which require the data be staged at a computing facility such as are brokered in Europe by consortia like PRACE. Data placement is reliant however on data accessibility, which is not simply based on the existence of the data in an accessible location, but is also based on the metadata associated with the core data that allows it to be correctly interpreted; it is based on the availability of services that understand that metadata and can so interact (and transport) the data with a minimum of manual configuration or direction.

Reductionism aside however, the key performance indicator used by most RIs is researcher productivityresearcher productivity. Can researchers use the RI to efficiently locate the data they need? Do they have access to all the support available for processing the data and conducting their experiments? Can they replicate the cited results of their peers using the facilities provided? This raises yet another question: how does the service provided to researchers translate to requirements on data placement and infrastructure availability?

This is key to intelligent placement of data—the of data—the existence of constraints that guide (semi-) autonomous services by conferring an understanding of the fundamental underlying context in which data placement occurs. The programming of infrastructure in order to support certain task workflows is a part of this.

Processing

The distribution of computation is a major concern for the optimisation of computational infrastructure for environmental science. Processing can be initiated on the behest of users, or can be part of the standard regime for data preparation and analysis embarked as part of the 'data pipeline' that runs through most environmental science research infrastructures. Given a dataset, an investigator can retrieve the data within to process on their own compute resources (ranging from a laptop or desktop to a private compute cluster), transfer the data onto a dedicated resource (such as a supercomputer for which they have leased time and capacity, Cloud infrastructure provisioned for the purpose, or for smaller tasks simply invoke a web service), or direct processing of the data on-site (generally only possible where the investigator has authority over the site in question, and generally limited to standard analyses that are part of the afore-mentioned data pipeline). Each of these options confers a (possibly zero) cost for data movement, data preparation, and process configuration.

...

Given constraints on compute capacity, network bandwidth, and quality of service, the most pertinent question in the sphere of optimisation is simply, given the sum of all activities engaged in by the research community at large, where should the data be processed?

It should be noted that the outputs of data processing are as much of concern as much as the inputs, especially if the curation of experimental results is considered within the purview of a given research infrastructure, and fold back into the domain of data curation.

Provenance

Good provenance is fundamental to optimisation—in optimisation---in order to be able to anticipate how data will be used by the community, and what infrastructure elements should be able conscripted to provide access to and processing capability over those data, it is necessary to understand as much about the data as possible. Thus provenance data is a key element of knowledge-augmented infrastructure, and provenance recording services are a major source of the knowledge that needs to be disseminated throughout the infrastructure in order to realise this ideal. Provenance is required to answer who, what, where, when, why and how regarding the orgins origins of data, and the role of an optimised RI is to rknow infer the answers for who, what, where, when, why and how regarding the each of those things as they regard the present and future use of those data. Ensuring that these questions can be asked and answered becomes more challenging the greater the heterogeneity of the data being handled by the RI.Quality checking of datasets , and so potential for runtime optimisation in particular will depend on the solutions for optimisation provided by the provenance task (T8.3) in ENVRI+.

As far as optimisation serving provenance in and of itself is concerned, the management of provenance data streams during data processing is the most likely area of focus. Preserving the link between data and their provenance metadata is also important, in order to ensure that the data is fit for purpose.particularly in cases where those metadata are not packaged with their corresponding datasets.

Curation

Streamlining the acquisition of data from data providers is important to many RIs, both to maximise the range and timeliness of datasets then made available to researchers, and to increase data security (by ensuring that it is properly curated with minimal delay, reducing the risk of data corruption or loss) is important.

In general, the principal concerns of curation are ensuring the accessibility and availability of research assets (especially, but not exclusively, data). High availability in particular requires effective replication procedures across multiple sites. It would be expedient to minimise the cost of synchronising replicas and to anticipate where user demand (for retrieval) is likely to be so as to minimise network congestion.

Cataloguing

Data catalogues are expected to be the main vector by which data is identified and requested by users, regardless of where that data is ultimately taken for processing and analysis. As such, the optimisation of both querying and data retrieval is of concern.

Identification and citation

With regard to identification and citation, it is necessary to ensure availability of identification services, and it is necessary to direct users to the best replicas of a given dataset that would ensure the most effective use of the underlying network.

Optimisation methodology

...