Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
The dashboard web application lets users manage their application containers: create, delete, start, and stop them as necessary.
Create several independent containers for various purposes, for example, using different base Docker images which provide different sets of pre-installed libraries and language tools.
Interactive server-side computational capabilities through Jupyter notebooks with Python, R, and MATLAB.
We provide Python and R libraries for communication with other SciServer components from the Jupyter notebooks running within SciServer Compute framework.
SciServer Compute User Interface
SciServer/Compute Architecture
· Homogeneous VM nodes serve as building blocks of a scalable system.
· Docker containers are used for process and resource isolation.
· A single identity service based on OpenStack Keystone provides seamless integration with other SciServer components.
· Direct access to large data archives and additional libraries and tools through shared, read-only data volumes.
· Personal persistent and scratch user space for long- and short-term storage of input data and results.
App Container
App Container
App Container
Docker
VM N
ode
Compute
Registry DB
SSO portal<username>
User containers
Data volume container
Shared containers
Keystone
Data archives
Scratch storage
User folders<username>
Images
Persistent storage
User folders<username>
Scratch storagevolume container
SciDrive
CasJobs
SkyServer
SciServer WebApps
Persistent storagevolume container
SciServer Compute
SciServer Compute uses Jupyter notebooks running within server-side Docker containers attached to large relational databases and file storage to bring advanced analysis capabilities close to the data. Apart from interactive notebooks in Python, R and MATLAB, SciServer Compute offers an API for running asynchronous tasks, also in Docker containers.
http://compute.sciserver.org
Running Asynchronous Tasks
Simple REST API for running asynchronous tasks inside the same container infrastructure that is used for running interactive Jupyter notebooks.
POST /dashboard/api/container HTTP/1.1Host: compute.sciserver.orgX-Auth-Token: 416aee7eab474...Content-Type: application/json
{"ImageName":"astro","Label":"test","Cmd":["python ~/workspace/persistent/test.py"],"Volumes":["sdss_das"]}
SciServerSciServer is a big data infrastructure project to develop a common environment for sharable computational research. SciServer is a fully integrated cyberinfrastructure system encompassing related tools and services. SciServer enables a new approach that will allow researchers to work with Terabytes or Petabytes of scientific data, without needing to download any large datasets. Some features we provide are· data storage for scientific users, tools for searching big datasets, and space for users to store and analyze their results.
· the ability to analyze data on our servers, keeping the computation close to the data to minimize data movement.
· access to query, analysis, and storage resources to researchers and educators and support for the longtail of science.
http://www.sciserver.org
For questions email us at [email protected] Funded by the U.S. National Science Foundation Award ACI–1261715
SciServer ComputeBringing Analysis Close to the Data
Jai Won Kim, Gerard Lemson Institute for Data Intensive Engineering and Science (IDIES), Johns Hopkins University
SciDrive
Compute
Login Portal
SkyServer
CasJobs
We o�er access to the complete SDSS dataset to the world through SkyServer. You can store any data you l ike in a private database in CasJobs, or as �les in SciDrive.
DataYou can �nd SDSS data with SQL queries through SkyServer or CasJobs.You can run queries through scripts using Compute.
Query
You can share your data and results with col leagues, and you can write �les in any format toSciDrive.
Collaboration
You can analyze the largest datasets with Python or R scr ipts through Compute.You can share scripts and results while Logged In.
Analysis
SkyQuery
You can cross-match your own data with SDSS data using our new SkyQuery tool .
Cross-Matching