Enterprise Research Infrastructure & Services

candlewhynotΔιαχείριση Δεδομένων

31 Ιαν 2013 (πριν από 4 χρόνια και 4 μήνες)

283 εμφανίσεις

Enterprise Research
Infrastructure & Services

Updates

Research Tech Lunch

April 28, 2011

DATABASES:

STRUCTURED

DATA SERVICES

Enterprise Research Infrastructure Systems

Allan Harris/DBA

ajharris@partners.org

(*Updated from December 2009*)

Overview


What database services does ERIS offer?


Current realms of DB management


HPC


DIPR

services


External DB support


Crimson


on track to move into
DIPR


PCPGM



in discussions to utilize Oracle
RAC

cluster



HPC

Solutions


hpcdb.research.partners.org


PostgreSQL
: 4 schemas


MySQL
: 17 schemas


Bringing Oracle and MS SQL to the clusters


Integration with
DIPR

DIPR

(
D
iscovery
I
nformatics
P
latform for
R
esearch)


PostgreSQL

(13 schemas)


MySQL

(69 schemas)


Oracle Database
RAC

(Real Application Clusters):


Version 11gR2, backwards compatibility available


26 user schemas in 4 instances


Polyserve

for MS SQL Server


Versions: 2005, 2008, 2008 R2


33 schemas in 6 instances


FileMaker Server Advanced


Version 11.0.2 (moving to 11.0.3)


31 hosted database files

Oracle RAC
(
R
eal
A
pplication
C
lusters)


Extensible option of Oracle Database


Oracle 11g Release 2


Single database accessed by multiple, coordinated
instances on multiple hosts


Leverages “Cache Fusion” over
Infiniband

interconnect for
SGA

unification


Advanced connectivity capabilities can spread single
service loads across multiple servers



Oracle RAC
(
cont’d
)

Oracle RAC
(
cont’d
)


Advantages


Built on commodity hardware


nodes can be dissimilar


Durability/scalability/flexibility


Cache Fusion


License consolidation


Implementation plans


Currently running on 2 (2 more available) HP BL685c AMD
-
based blade servers


Each node has 16 processor cores and 64GB RAM


Can expand/contract as necessary


PolyServe

for Microsoft SQL Server


NOT a replacement of TSO services


Consolidation


Ease of deployment and patching


Simplified management


Multi
-
instance access to RO datasets


CITED

PolyServe

for SQL
(cont’d)

FileMaker

Server Advanced


Hosted FileMaker solution


Exposes FileMaker databases for enterprise
consumption


Instant Web Publishing/
XSLT
/
PHP

Resources


http://research.partners.org




http://h18000.www1.hp.com/products/storage/software/polyser
ve/db_utility/sql/index.html




http://www.oracle.com/technology/products/database/clusterin
g/pdf/twp_rac11gR2.pdf


Applications for HPC Linux/Windows



Schedulers


LSF / PBS / Torque (Maui) …



GENOMICS / SEQUENCING :
BioScope

/
PicardTools

/ SAM BAM TOOLS /
TopHat

/
CuffLinks

/ CASAVA / BWA …



STATISTICS : R, SAS, Octave …



Special : SSE and MPI


0.00E+00
5.00E+04
1.00E+05
1.50E+05
2.00E+05
2.50E+05
3.00E+05
3.50E+05
2007 Q3
2007 Q4
2008 Q1
2008 Q2
2008 Q3
2008 Q4
2009 Q1
2009 Q2
2009 Q3
2009 Q4
2010 Q1
2010 Q2
2010 Q3
2010 Q4
Used Hours

0
2000
4000
6000
8000
10000
12000
14000
16000
18000
2009
Q1
2009
Q2
2009
Q3
2009
Q4
2010
Q1
2010
Q2
2010
Q3
2010
Q4
Hours

Quarter

Windows Cluster Usage Hours

General Domain

Genomics and NextGen Sequencing

Proteomic and protein analysis

medical and physiological simulation

Biostats and Statistical Genetics

Image Analysis

i2b2/RPDR

Text Mining

Other

total

Q1 11
(total)

43

25

13

106

57

7

10

32

293

Total Number of Users: ~ 500 / Number of Cores ~ 800



16 HP BL460c G1 Blades


8 Cores (2x Xeon 5450 @ 3ghz)


64GB Memory



EVA 8100 SAN Storage



VMWare ESXi 4.1











271 Virtual machines hosted


339.2/1024GB Memory in use


6.54TB provisioned/3.56TB utilized



Expanded each VM host memory from 32 to
64GB


New builds managed by Spacewalk


Standard build is now CentOS 5


All new VMs are 64 bit



Groups can request, free of charge, either:


2 “Small” VMs


512MB memory


8GB Disk


1 “Medium” VM


1GB memory


16GB Disk


Standard OS build or Custom OS of choice.


CentOS 5, 64 Bit


User authentication with Partners AD


We manage OS updates


Nagios monitoring


SSH access with local sudo


Standard package software installed upon
request


User has access to yum for package installs



Any VMWare compatible OS can be installed


User is responsible for security updates


User is responsible for software installs


User must provide OS media/license


Single local admin account configured with
remote access


SSH for Linux, RDP for windows


Research Web Proxy


Can proxy http/https traffic to DIPR VMs


Running apache mod_proxy


*.partners.org wildcard cert installed for SSL



CentOS 6 will be available once
released/tested


Existing VMs will be converted to CentOS


Additional standard build configurations

EDC/Survey


3 options EDC


REDCap
,
Velos
,
StudyTrax


2 options Surveys


REDCap
,
LimeSurvey


REDCap


Went Live May 2010


Today:


6
th

in PRODUCTION projects


2
nd

in DEVELOPMENT projects


2
nd

in TOTAL projects


2
nd

in TOTAL USERS



Projects Report















REDCap

BIDMC

CHB

Joslin

PHS

Total

%

Development

21

58

1

152

232

63%

Production

7

43

3

81

134

37%

Inactive / Archived

0

0

0

0

0

0%

Total

28

101

4

233

366

100%













REDCap Survey

BIDMC

CHB

Joslin

PHS

Total

%



Development

13

28

1

303

345

60%



Production

11

14

0

130

155

27%



Inactive / Archived

1

5

0

69

75

13%



Total

25

47

1

502

575

100%















Total Projects

BIDMC

CHB

Joslin

PHS

Total

%



Development

34

86

2

455

577

61%



Production

18

57

3

211

289

31%



Inactive / Archived

1

5

0

69

75

8%



Grand Total

53

148

5

735

941

100%

















Info Sessions/Meetings

BIDMC

CHB

Joslin

PHS

Harvard

Total



Info Sessions/Meetings

69

50

5

166

12

302



Result in Project Lead

39

34

3

^

2

78



No Resolution at this time

24

10

2

^

6

42

Chose Alternative Solution

6

6

0

6

0

18

Avg Time/Session:

107 min

79 min

100 min

-

67 min

92 min

-

metric not tracked,

^ participants provide general feedback after presentations

Trainings

BIDMC

CHB

Joslin

PHS

Harvard

Total

Basic Programming Classes

*

*

*

13

*

13



Number of Attendees

*

*

*

69

*

69

* trainings are included in the stats for Consultations (contacts > 15 mins)















Support

BIDMC

CHB

Joslin

PHS

Harvard

Total



Contacts
(< 15 mins; emails, phone calls)

83

94

11

1826

**

2014




Consultations
(contacts > 15 mins)

146

171

13


197

3**

530



Avg Time/Consultation

46 min

45 min

60 min

-

37 min

47 min



** Harvard support included in PHS support statistics

-

metric not tracked

Future Features

21CFR Part 11



Merging REDCap + REDCap
-
Survey


External Systems Interoperability


REDCap Unplugged (No Internet)


Improved Language Rendering


Matrix Question Type / Table Format




29

Interesting Use
Cases

30

In development / production:


Student/Class, Program Evaluations


Longitudinal Survey Studies


Project Management Calendar


Study Bio
-
specimen Tracking/Scheduling Checklist




… So many possibilities!

Frequently Asked
Questions (FAQ)

31


Q: How much training is required to
use and design in REDCap?

A: Minimal training is needed. Development
support available in the form of:



Online Tutorials and Videos



Periodic Training



Refined User Guides



Direct phone/email contact with

support staff


Q: How much experience with
programming, networking and/or
database construction is required to
use and design EDC tools in REDCap?

A: No programming, networking or database
experience required.

Use Point
-
and
-
Click interface to design your
EDC tool(s).

Contacts

Contact Information for Harvard Catalyst REDCap Support Staff


Lynn Simpson

MGH, BWH & McLean

edcsupport@partners.org

Phone: 617.643.7711

http://rc.partners.org/edcredcap



Chris Botte

BIDMC, Children’s & Joslin

edc@bidmc.harvard.edu

Phone: 617.754.8828


For support for other Harvard schools or affiliated academic health care centers,

contact either Lynn or Chris.

32