KAREN2 Background
There are a number of diverse factors that are driving the development of KAREN2.
On one side there are strategic issues such as the evolution in the manner in which research is being carried out, eg data intensive science becoming more pervasive, the rapidly increasing use of “research based cloud computing facilities”, and the increasing emphasis on collaborative research and strong organic growth in the use of current network bandwidth. On the other side are pragmatic issues related to existing supply contracts, procurement timeframes and future access to increased international bandwidth (with Pacific Fibre).
Data and research group size
Research in many areas is undergoing a fundamental change with significant growth in data driven research. The increasing sophistication of instrumentation coupled with the availability of low cost high performance computing resources is producing a “tsunami” of data.
The relative small size of research groups, particularly in New Zealand, means that such groups often have poor access to the necessary compute and storage resources. The large scale computing resources demanded by this type of research activity is increasingly moving out of the traditional data centre of our research organisations and into the cloud. The move by funding organisations, such as MSI, to make data more accessible so that it may be re-used will stimulate further growth in the use of the network.
The cloud
In New Zealand two recent developments, New Zealand Genomics Ltd (NZGL) and the National eScience Initiative (NeSI) propose to provide extensive cloud based infrastructures for data storage and processing. These initiatives will significantly increase the amount of traffic on the network. In addition many of our members now use cloud-based resources to deliver some of their line-of-business applications, eg the use of Google email services by some Universities for their entire student population and staff.
The move to cloud-based systems is likely to continue as institutions strive to manage their cost structures. These types of change will see a concentration of research infrastructure in fewer places but with the expectation of quality access from everywhere. Such specialist facilities will start to demand increasing flexibility from the network. For example, geographically disparate nodes within the NeSI super-computer environment may require direct interconnects (potentially individual light paths) so that they can inter-operate without the need to traverse the wider network infrastructure. The very large storage requirements of NZGL will challenge conventional data backup approaches and will give rise to requirements for real-time remote backup across KAREN2.
These increases in data generating capacity with its attendant need to transfer very large files, coupled with the growth in cloud computing will be the fundamental driver for access to increasing network bandwidth. This trend is already well established internationally.
Existing contracts, procurement timeframes and future access to bandwidth
Even without the changes referred to above, utilisation of the network is growing strongly and based on projections congestion will start to be experienced on some links during 2012. The current National Network contract expires in December 2013 and while we might expect to be able to renegotiate a supply contract beyond that date, without the move to a “dark fibre” based transport layer, REANNZ will not have a cost effective way to develop the bandwidth headroom and connection flexibility that is required of an advanced network.
The process to design, procure and implement KAREN2, based on dark fibre infrastructure, has an 18-24 month planning horizon and a number of key activities need to be progressed now. The advent of 40Gb/s international bandwidth from Pacific Fibre in mid-2014 also provides significant motivation to get this project underway.