Tuesday, December 09, 2008

Virtual Machine Infrastructure for NZ Hospitals

The New Zeeland West Coast District Health Board has issued a request for proposal for a Virtual Machine Infrastructure to replace its current 41 are physical servers. This is an interesting example of an attempt to rationalise a complex computer system. The board will need to decide if its two server rooms (primary and backup) are sufficient and how if more than just two physical servers are needed. The board would need to balance the saving in hardware and energy (and lower greenhouse gas emissions) this would provide against the security are reliability issues.
GETS Reference: 24453
Title: New Zealand based opportunityVirtual Machine Infrastructure
Request for Proposal RFP08/01

General Information:

WCDHB is the District Health Board that serves the health needs of the West Coast of the South Island. It currently has three hospital sites in Greymouth, Westport and Reefton.

It also has a health clinic in Hokitika and many smaller satellite based clinics up and down the Coast. WCDHB services an area from Karamea in the north to Haast in the south.

WCDHB has a wide area network spanning Karamea to Fox Glacier, including all major town centres on the West Coast.

WCDHB has 55 physical servers, of which 41 are physical servers and 14 virtual servers using VMware Infrastructure edition on a single physical server, using local disk as storage.

The servers are split between two server rooms, a primary and a backup (most being in the primary room), with a 4 Gig fibre backbone running between them.

The purpose of this Request for Proposal (“RFP”) is to invite external companies to submit their proposals to WCDHB with information on their skills, services and experience in providing Virtual Machine Infrastructure services and products.

The information is requested so that WCDHB can:
  • Identify organisations interested in and capable of delivering these products/services; and
  • Identify different methods of providing such products/services and a preferred solution/product.
Following the evaluation of the RFP responses, WCDHB may:
  • Enter into negotiations with preferred supplier(s); and/or
  • Conclude the process without awarding any contracts.

Site visits or workshops if needed: Available during 8th to 12th of December.

To access the RFP documentation please download from ...

Additional Documentation to Download... WCDHB VM RFP08 01.doc RFP documentation WORD 299.5kb
RFP08 01 Questions and Answers 09 Dec 08.doc Q & A's # 1 - Dated 9 December 08 WORD 190kb

Relates to the following TenderWatch Categories
841 Project management relating to IT service and delivery
842 Software implementation services
453 Computer software
849 Other computer services

Labels: , , , ,

Tuesday, April 29, 2008

End-user Computing at the Australian Taxation Office

The Australian Taxation Office has issued a request for expressions of interest for End-user Computing Services (that is desktop PCs, laptops and mobile devices. A 27 page overview of the Current Environment at the ATO is provided. This gives a good overview of the typical computing setup in a large government agency.
End-user Computing Services for the Australian Taxation Office
ATM ID EOI 08.063
Agency Australian Taxation Office
Category 43210000 - Computer Equipment and Accessories
Close Date & Time 26-May-2008 2:00 pm (ACT Local time)
Show close time for other time zones
Publish Date 28-Apr-2008
Location ACT
ATM Type Expression of Interest
See ATM Documents

The ATO aims to virtualise:
      1. Data Centre Virtualisation

      2. This project aims to virtualise the majority of the Tax Office application server infrastructure (development, test and production environments). Although the current project scope does not cover the existing EUC infrastructure, this may change with the Remote Access project, which is likely to have a dependency on some form of virtualised infrastructure in the EUC architecture.

      3. Additionally the testing environment for EUC end-users is being virtualised. This consists of a combination of racked workstations and virtualised workstations. This solution is currently evolving but is expected to have an impact on the EUC architecture.

      4. National office development workstations have been virtualised within the desktop box.

      5. There are plans to improve on the developer design and extend to all developer workstations across Australia.

But there doesn't seem to be any intention to replace the desktop computers with thin clients, or to made use of handheld mobile devices.

Labels: ,

Thursday, April 24, 2008

Scheduling in heterogeneous virtualised clusters

With all the talk of "virtualisation" of servers to rationalize computer use and reduce energy consumption, it is worth keeping in mind that getting applications to run well on such systems can be difficult. There is a seminar, called at short notice, in Canberra, about this today:

Seminar Announcement

Department of Computer Science, FEIT
The Australian National University

Date: Thursday, 24 April 2008
Time: 4:00 pm to 5:00 pm
Venue: Room N101, CSIT Building [108]

Speaker: Muhammad Atif

Title: Scheduling in heterogeneous virtualised clusters
(Thesis Proposal Review seminar)


Clusters of commodity processors are highly heterogeneous in nature. The volume of computation performed by one compute node in the cluster can be drastically different from other nodes due to architectural or communication interconnect differences. The conventional scheduling
solutions do not take detailed heterogeneity aspects into consideration,
and thus are not very effective.

Virtualisation can be utilized to normalize the heterogeneity of a
cluster to some extent. With growing maturity and hardware support, the
performance gap between native and virtualised operating systems is
decreasing making it a good candidate for high performance computing

The concept of live migration of operating systems can be utilized to
maximize the throughput of the compute farm and the turnaround times of
the submitted jobs. This can be done by developing a run-time job
estimation and remapping service capable of determining more optimized
hardware environment for a job at runtime, resulting in high throughput
and improved turnaround times in the compute cluster.

This seminar will discuss the prospects of using virtualisation in HPC
and some proposed solutions. we will also present some early results of
our experiments with virtualised compute clusters and discuss the short-
comings of current Virtualisation solutions with HPC viewpoint.


Muhammad Atif has done Masters in Software Engineering from the National University of Sciences and Technology (NUST), Pakistan. In 2003, he was was awarded the prestigious president's gold medal for his excellent academic record at his university.

He is currently pursuing PhD in the field of Scheduling in High
performance computing from from the Department of Computer Science,
ANU. His research interest includes virtualization of operating systems
and performance estimation in cluster computing. His past research
interests include natural language processing and anomaly detection in
agent based computing.

URL: http://cs.anu.edu.au/lib/seminars/seminars08/dept20080424

Labels: ,

Tuesday, April 15, 2008

Geoscience Australia has issued a Request for Tender for Windows and Linux Servers. They are going to use virtualization to rationalize their server hardware. However, the ability to monitor power use and have power reduction features does not seem to be mentioned:
8.4 Technical Specifications
8.4.1 Geoscience Australia is seeking to establish a contract for the supply of a range of business capable servers. The following details the specifications required of the servers:

8.4.2 Capable of operating the following operating systems as a minimum

a) Windows 2000 server
b) Windows 2003 server both 32 and 64 bit
c) Windows 2008 server both 32 and 64 bit
d) Red Hat Linux enterprise
e) Capable of multiple core/multi processor operation
f) Have a three year warranty on all parts and labour
g) Minimum 4 hour (24*7) warranty and extendable for a fourth and fifth
h) Fit into a standard 900mm computer rack
i) Have redundant power supplies
j) Dual native 10/100/1000 Mb/sec NIC capability supporting PXE
k) Capable of Raid 0,1 and 5 as a minimum
l) Have an external scaleable disk solution
m) Capable of supporting a multiple range of HDD including SAS, SATA etc with choice of AMD or Intel CPU acceptable
n) A capability for the addition of external disk storage
o) Capable of SCSi attached tape storage
p) Have a DVD-ROM
q) Have the capability of either PS/2, USB or IP based KVM
r) Have an internal PCI (minimum) expansion capability, including the
ability to handle 3D graphics cards
s) Have redundant fans capable of cooling at the extreme operations of the
t) Be supported by a software management system than provides remote
hardware status and alerting, diagnostics and logging and firmware/BIOS
u) Must be capable of email fault alerting
v) Supports VMW are V3x virtualisation
w) Have a RAM expansion capability between 2Gb and 64Gb
x) Have a FSB of at least 1066Mhz
y) Be capable of Cache expansion greater than 6Mb
z) Be capable of supporting Fibre connectivity to 10Gb
aa) Be capable of provisioning monitoring, alerting and independent
actioning of environmental information
8.5 First Year Requirements

From: Request for Tender for the Provision of Computing Equipment to Support Geoscience Australia's Windows/Linux Server Environment, ATM ID RFT2008/1106, Geoscience Australia, 14-Apr-2008

Labels: ,