![Page 1: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/1.jpg)
Anushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant
GALAXY COMMUNITY CONFERENCE 2014
![Page 2: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/2.jpg)
• Who We Are
• SlipStream Galaxy Appliance
• Science vs Infrastructure
• Hybrid Computing: Flexibility and Scale
• Looking Ahead
![Page 3: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/3.jpg)
• Staffed by scientists forced to learn IT to get research done
• Served over 400 organizations • Academic, Non-profit • Government, Military • Pharm, AgBio, Biotech • Cloud & Datacenter
Providers
Over a Decade of Life Sciences IT Consulting
![Page 4: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/4.jpg)
Bridging the IT Gap
• Encapsulate IT best-practices expertise to eliminate redundant effort spent building IT systems and installing software
• Reduce the barrier to entry into data analysis by
improving accessibility of the Galaxy platform
OFFICIAL APPLIANCE PROVIDER FOR THE GALAXY PROJECT
![Page 5: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/5.jpg)
Powerful dedicated desktop server pre-‐configured with a fully operational production instance of Galaxy
![Page 6: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/6.jpg)
Scientific Goals
IT Challenges
IT & informatics support
Disconnected resources
Infrastructure management
Diverse computing needs
Focus on the science
Leverage available resources
Start small and scale
Use a variety of analysis tools
![Page 7: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/7.jpg)
The Problem • Enable users to utilize additional resources
available to them beyond those in SlipStream Galaxy Appliance • Local resources • Cloud resources
![Page 8: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/8.jpg)
The Goal
• Make SlipStream Galaxy a central gateway to additional resources
• Keep things simple
![Page 9: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/9.jpg)
What We Did (Example 1)
• Customer wants to leverage existing SGE environment and resources
• Jobs should spill over once the appliance is “busy”
• Solution: Cross-mount storage and implement transfer queue
![Page 10: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/10.jpg)
SGE
SGE
slipstream_queue
transfer.q galaxy.q
other.q …
Customer “Rocks” Cluster Customer SlipStream Appliance
Disk CPU CPU NFS Disk
![Page 11: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/11.jpg)
SGE
SGE
slipstream_queue
transfer.q galaxy.q
other.q …
Customer “Rocks” Cluster Customer SlipStream Appliance
Disk CPU CPU NFS Disk
Life is Good…
![Page 12: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/12.jpg)
SGE
SGE
slipstream_queue
transfer.q galaxy.q
other.q …
Customer “Rocks” Cluster Customer SlipStream Appliance
Disk CPU CPU NFS Disk
Need more resources!
![Page 13: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/13.jpg)
SGE
SGE
slipstream_queue
transfer.q galaxy.q
other.q …
Customer “Rocks” Cluster Customer SlipStream Appliance
Disk CPU CPU NFS Disk
Load sensor trips, jobs start to transfer to the cluster
NFS via autofs
Hooks… clusterload.sh transfer_resume.sh transfer_starter.sh transfer_suspend.sh transfer_terminate.sh
![Page 14: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/14.jpg)
SGE
SGE
slipstream_queue
transfer.q galaxy.q
other.q …
Customer “Rocks” Cluster Customer SlipStream Appliance
Disk CPU CPU NFS Disk
Life is better!
NFS via autofs
![Page 15: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/15.jpg)
SGE
SGE
slipstream_queue
transfer.q galaxy.q
other.q …
Customer “Rocks” Cluster Customer SlipStream Appliance
Disk CPU CPU NFS Disk
Maximum number of jobs transferred reached… need more resources? Now what?
NFS via autofs
X Jobs are now ‘qw’!
![Page 16: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/16.jpg)
![Page 17: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/17.jpg)
SGE
SGE
slipstream_queue
transfer.q galaxy.q
other.q …
Modified StarCluster AMI Customer SlipStream Appliance
Disk CPU CPU NFS EBS
&
NFS sync
Hooks… clusterload.sh transfer_resume.sh transfer_starter.sh transfer_suspend.sh transfer_terminate.sh
![Page 18: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/18.jpg)
• Star Cluster has a powerful decision engine
• Leave the head node running and StarCluster will automatically spin up and spin down workers depending on level of “bursting” happening.
• Configuration management with transfer queue scripts to provision head node automatically…
![Page 19: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/19.jpg)
The Challenges with Bursting
• Users want to run jobs from the CLI as well as Galaxy that use external resources
• Tool compatibility in heterogeneous environments (Rocks cluster is RHEL, Appliance is Ubuntu)
• Share storage between resources
• Again, keep things simple
![Page 20: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/20.jpg)
How We Solved Them
• Users want to run jobs from the CLI as well as Galaxy that can use external resources
• By using a transfer queue, users interact with SGE in a familiar way from the CLI as well as through Galaxy
![Page 21: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/21.jpg)
How We Solved Them • Tool compatibility in heterogeneous
environments (Rocks cluster is RHEL, Appliance is Ubuntu)
• Some toolshed tools with precompiled binaries are designed to be compatible between RHEL/CentOS & Ubuntu
• Others must be built carefully • StarCluster is Ubuntu-based so it is easier to
maintain tool compatibility
![Page 22: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/22.jpg)
How We Solved Them
• Share storage between resources
• NFS solution was the customer’s preferred starting point
• Delivering NFS in cloud could benefit using asynchronous caching (ex. Avere)
![Page 23: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/23.jpg)
How We Solved Them
• Again, keep things simple.
• Using existing SGE environment, NFS, etc. minimizes integration effort
• Customer’s interaction with appliance from the CLI and Galaxy doesn’t change
![Page 24: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/24.jpg)
Good start… but how do we expand this concept in the future?
• Light Weight Runner to increase abstraction for
resource aware scheduling
• Docker/LXC to provide isolation and portability of tools
• Apache Mesos for resource aware meta-scheduling
![Page 25: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/25.jpg)
Scientific Goals
IT Solutions
Appliance and Galaxy are fully supported
Central gateway to integrated resources
Scale infrastructure as-‐needed
Launch tools on appropriate hardware
Focus on the science
Leverage available resources
Start small and scale
Use plethora of analysis tools
![Page 26: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/26.jpg)
New Reference Design Collaboration with SGI and Intel to provide an
even more powerful, affordable appliance
Enhanced Galaxy Support Partnership with BioStar Genomics to develop additional Galaxy support and service offerings
Scalable Infrastructure
Continue to build infrastructure integrations that dynamically fit scientific computing needs
![Page 27: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/27.jpg)
SlipStream Appliance: Galaxy Edition A high performance solution for data analysis
Why SlipStream Galaxy
• 10+ years of Life Science IT expertise • Dedicated, flexible, scalable resource
• Infrastructure and Galaxy administration support
![Page 28: Anushka Brownley, Senior Scientific Consultant Aaron ... · PDF fileAnushka Brownley, Senior Scientific Consultant Aaron Gardner, Senior Scientific Consultant GALAXY COMMUNITY CONFERENCE](https://reader031.vdocuments.mx/reader031/viewer/2022030419/5aa617007f8b9a185d8e2ebe/html5/thumbnails/28.jpg)
ONGOING EARLY ACCESS PROGRAM (Limited Availability)
Visit the BioTeam booth for more information!!
www.bioteam.net/slipstream/galaxy-edition