Research Infrastructure

RI Update 2015 August

Since April, we have added almost one million pounds worth of new hardware to the University Research IT infrastructure. We expand on how the money was spent — what new resources are available — below. For more information, please contact the Research Infrastructure Team at

its-ri-team@manchester.ac.uk

Relocation of platforms from the Kilburn Data Centre to Joule House

As many users of our systems will know, we have recently been moving some of our computational platforms from the dated Kilburn Data Centre to a new data centre at Joule House, at Salford Quays. The moves of Hydra, Redqueen, Zrek (zCSF) and Incline (iCSF) have all been completed successfully, with only minor glitches. The new data centre is a much better environment for our hardware which tends to “run hot” and ensures there is plenty of capacity for future expansion.

End-of-Financial-Year Rush

The end-of-year procurement rush is now thankfully over! It has been a record summer for financial contributions to our infrastructure:

  • almost 200k to the CSF and
  • nearly 175k to Hydra;
  • also, 400k for Isilon storage from MRC-funded projects.

The total academic contribution to University Research IT infrastructure has now reached approximately 3.5 million pounds!

CSF Update

In the last six months or so we have added over 1000 cores of the latest Intel Haswell CPUs to the system. Each node has 24 cores, 128 GB RAM and 1.6 TB of local “tmp” space; all of these nodes are Infiniband-connected.

The CSF now comprises over 7000 cores.

RDS (aka Isilon) news

A further expansion of capacity in this service is in progress.

The Research Data Service (aka Isilon) offers resilient storage — all data is replicated and snapshotted. (For more info visit the RDS Web site.) The amount of data stored grew over the last year at around 40 TB per month — the total is now nearly 800 TB.

Once the current expansion is completed, We expect another 500 TB of storage to be made available to users.

Hydra Update

New nodes — by the time you read this, another 10 nodes, each with 512 GB RAM, will be in production, doubling the size of the cluster. The available local, high-performance scratch space will will have increased both in terms of capacity and IO bandwidth too.

Significant, further developments related to Hydra are coming. Watch this space!

Dedicated Research VM Infrastructure

The long-planned, dedicated Research VM Infrastructure will be in production very soon. This platform will run on the same sustainable financial model as the CSF, based on academic contributions of funds — contributors receive VM resources equivalent to the funds contributed.

  • We expect to be able to offer VMs of a much higher specification than those on the current/pilot Research VM Service.
  • We will exploring the possibility of offering a visualization service via these VMs — the VM hosts include Nvidia K2 GPUs.

Comments are closed.