I have just had another great week in Tokyo last week with the Global Grid Forum 17^th Symposium combined with the GridWorld Japan event. Whilst the conference & expo clearly demonstrated the interest of the Japanese audience in Grid-related technologies and solutions, Grid experts from academia & private R&D labs around the world continued debating the next steps to accelerate the adoption of standardised Grid solutions, beyond scientific applications environments where the focus on compute power of Grid prevails and more towards the Enterprise where many resource management requirements co-exist across multiple IT infrastructure layers.

I was also encouraged to see the Enterprise Grid Alliance troops joining the GGF contingents as a prelude to the likely GGF/EGA merger, bringing in the new organization community some extra resource and efforts to bring Grid into the data centre. From a data storage perspective, this is also good news since EGA has also been working on Data Provisioning requirements for Enterprise Grid, in collaboration with the SNIA. Let’s face it: all data centres are (also) in the business of managing the availability of Enterprise Information for business applications and users, so if Grids want to succeed as becoming this new generation of IT infrastructure, it’d better address Enterprise data & storage resource requirements too. In fact, many early cases of Grid deployments show that data access & storage resource management rapidly  emerge as the next topics to tackle.

Of course the need to address data management requirements such as data movement, data replication, security or naming conventions are being looked at in GGF, especially in the Data Area of GGF. The EGA forces will certainly help extending GGF’s efforts to include Enterprise requirements to efficiently manage IT resource in Grid with measurable technical and business objectives. If Grid is about managing services, leveraging web services architecture, the inclusion of data services will inevitably reach out to Enterprise storage solutions. To a very large extent, today’s Enterprise storage solutions are already handling many of the management tasks for data movement, replication or the virtualization of data
containers. Since Grids won’t appear by magic in the data centre overnight, Grid-like technologies will need to be gradually introduced in a compatible fashion with the incumbent data storage solutions, that is to say including some standardized management interface and data access protocols. This is why several months ago my colleague Ken Wood and I helped SNIA starting a Grid-focused discussion group along with several other interested SNIA members in order to determine the role of the storage industry in Grid, in collaboration with Grid industry groups such as EGA and GGF. Some of our thoughts are coming to fruition and we wanted to start sharing them with the Storage Networking Community Group of GGF that has just been formed last week in Japan.

This group of experts sees two main trends in the “Storage and Grid” discussion that could possibly merge in the future:

1) The first one concerns the readiness and gaps of existing data storage solutions and standards to be utilized / consumed by today’s (emerging) Grids, data or compute centric. This is an area where both Grid and Storage architectures remained aminly customized today and where many standard and architecture alignments work has to happen with vendors and standards organizations. We call it “Storage for Grid”.

2) The use of Grid architectures to provide storage solutions and services. Most Grid environments are meant to “globally” schedule the use of IT resource through a set of (standardized) services…This should certainly apply to Enterprise data storage. We call it “Storage Grid” or “Grid-based Storage”.

Even if the views of Grid researchers and Enterprise storage solution providers have yet to be reconciled, I am encouraged by the willingness to work together and to embrace the numerous Storage and Grid challenges at GGF, EGA and SNIA.

Comments RSS