Cloud Computing for Science

Transcription

1 Cloud Computing for Science June st International Conference on Scientific and Statistical Database Management Kate Keahey Nimbus project lead University of Chicago Argonne National Laboratory

2 Cloud Computing is in the news is it good news for Science?

3 Cloud Computing for Science Complex codes Need for control

4 Grid Computing Assumption: control over the manner in which resources are used stays with the site Site A R R R Site B R R R VO-A Site-specific environment and mode of access Site-driven prioritization But: site control -> rapid adoption

5 Cloud Computing Change of assumption: control over the resource is turned over to the user Site A R R R Site B R R R VO-A Enabling factors: virtualization and isolation Challenges our notion of a site Lends itself to more explicit service level negotiation But: slow adoption

6 Grids to Clouds: a Personal Perspective First STAR production run on EC2 Xen released EC2 goes online Nimbus Cloud comes online A Case for Grid Computing on VMs In-Vigo, VIOLIN, DVEs, Dynamic accounts Policy-driven negotiation First WSRF Workspace Service release EC2 gateway available Context Broker release Support for EC2 interfaces

7 Benefits to Consumers Eliminate expense of acquiring, managing and operating hardware capital expense 6/5/09 Elastic computing Pay-as-you-go model operational expense The Nimbus Toolkit: http//workspace.globus.org

8 Benefits to Providers Economies of scale to amortize the costs of buying and operating resources Avoid cost and complexity of managing multiple customer-specific environments and applications Streamline and specialize

9 Unclouding the Cloud Software-as-a-Service (SaaS) Community-specific applications and portals Platform-as-a-Service (PaaS) Infrastructure-as-a-Service (IaaS) 6/5/09 The Nimbus Toolkit: http//workspace.globus.org

10 IaaS Infrastructure

11 Nimbus: Cloud Computing Software Nimbus goals: Allow providers to build clouds Private clouds (privacy, expense considerations) Workspace Service: open source EC2 implementation Allow users to use cloud computing Do whatever it takes to enable scientists to use IaaS Context Broker: turnkey virtual clusters Allow developers to experiment with Nimbus For research or usability/performance improvements Community extensions and contributions

12 The Workspace Service VWS Service

13 The Workspace Service The workspace service publishes information about each workspace Users can find out information about their workspace (e.g. what IP the workspace was bound to) VWS Service Users can interact directly with their workspaces the same way the would with a physical machine.

14 Cloud Computing Ecosystem Appliance Providers Marketplaces, commercial providers, Virtual Organizations Appliance management software Deployment Orchestrator VMM/DataCenter/IaaS User Environments VMM/DataCenter/IaaS User Environments

15 Turnkey Virtual Clusters IP1 HK1 IP2 HK2 IP3 HK3 IP1 HK1 IP1 HK1 IP1 HK1 IP2 HK2 IP2 MPI HK2 IP2 HK2 IP3 HK3 IP3 HK3 IP3 HK3 Context Broker Turnkey, tightly-coupled cluster Shared trust/security context Shared configuration/context information

16 What s in the Box? context broker EC2 WSRF storage service workspace service IaaS gateway workspace resource manager workspace pilot workspace control EC2 potentially other providers context client cloud client workspace client

17 Open Source IaaS Implementations OpenNebula Open source datacenter implementation University of Madrid, I. Llorente & team, 03/2008 Eucalyptus Open source implementation of EC2 UCSB, R. Wolski & team, 06/2008 Cloud-enabled Nimrod-G Open source implementation of EC2 Monash University, MeSsAGE Lab, 01/2009 Industry efforts openqrm, Enomalism

18 Nimbus: Extensions Nimbus core: Tim Freeman & David LaBissoniere (UC/ANL team) Nimbus monitoring: Ian Gable & team (UVIC ATLAS) Cumulus: Raj Kettimuthu and John Bresnahan (ANL) EBS: Marlon Pierce, Xiaoming Gao, Mike Lowe (IU) Others: OpenNebula project (University of Madrid) Descher et al (Technical U of Vienna): privacy extensions

19 Scientific Cloud Resources and Applications

20 Science Clouds Goals Enable experimentation with IaaS Evolve software in response to user needs Exploration of cloud interoperability issues Participants University of Chicago (since 03/08, 16 s), University of Florida (05/08, s, access via VPN), Masaryk University, Brno, Czech Republic (08/08), Purdue (09/08) In progress: Grid5K, IU, Vrije Using EC2 for large runs Science Clouds Marketplace: OSG cluster, Hadoop, etc. Come and run:

21 Who Runs on Nimbus? 100+ DNs projects ranging across Science, CS, education, build&test

22 STAR experiment STAR: a nuclear physics experiment at Brookhaven National Laboratory Studies fundamental properties of nuclear matter Problem: computations require complex and consistently configured environments that are hard to find in existing grids

23 STAR Virtual Clusters Virtual resources A virtual OSG STAR cluster: OSG head (gridmapfiles, host certificates, NFS, Torque), worker s: SL4 + STAR One-click virtual cluster deployment via Nimbus Context Broker From Science Clouds to EC2 runs Running production codes since 2007 Work by Jerome Lauret, Leve Hajdu, Lidia Didenko (BNL), Doug Olson (LBNL) The Quark Matter run: producing just-in-time results for a conference:

24 STAR Quark Matter Run Gateway/ Context Broker Infrastructure-as-a-Service

25 STAR Quark Matter Run (2) Application stats: Processed 1.2 M events Moved ~1TB of data over duration (small I/O needs) Run facts: 300+ s over ~10 days Instances, 32-bit, 1.7 GB memory: Cost: EC2 default: 1 EC2 CPU unit High-CPU Medium Instances: 5 EC2 CPU units (2 cores) Comp: ~ $6,000: ~ $1,7 K (default) + ~ $3,9K (medium) Data: ~ $150

26 A Large Ion Collider Experiment (ALICE) Heavy ion simulations at CERN Problem: integrate elastic computing into current infrastructure Collaboration with CernVM project With Artem Harutyunyan and Predrag Buncic

27 Elastic Provisioning for ALICE HEP ALICE queue queue sensor AliEn Context Broker Infrastructure-as-a-Service

28 Alice HEP Experiment at CERN CHEP09 paper, Harutyunyan et al. Can we elastically extend a local scheduler?

29 Sky Computing Change of assumption: we can now trust remote resources Site A R R R Site B R R R VO-A Enabling factors: cloud computing and virtual networks Network leases would help Instead of a bunch of disonnected domains, one domain overlapping the Internet

30 Sky Computing Environment U of Chicago U of Florida ViNE router ViNE router ViNE router Purdue Work by A. Matsunaga, M. Tsugawa, University of Florida

31 Hadoop in the Science Clouds U of Chicago U of Florida Hadoop cloud Purdue Papers: Sky Computing, by K. Keahey, A. Matsunaga, M. Tsugawa, J. Fortes. Submitted to IEEE Internet Computing. CloudBLAST: Combining MapReduce and Virtualization on Distributed Resources for Bioinformatics Applications by A. Matsunaga, M. Tsugawa and J. Fortes. escience 2008.

32 Cloud Computing for Science: Issues and Challenges

33 Building the Ecosystem Configuring and maintaining appliances Not just VMs, a variety of formats CernVM, rbuilder (rpath) Licenses Still vendor-specific approaches Getting used to dynamic sites Host certificates and keys, community visibility, failure processing, etc. Infrastructure and leveraging

34 Security and Privacy Issues Leaks in isolation, improper use, new technology issues Lack of features Fine-grained authorization Paper: Palankar et al., Amazon S3 for Science Grids: a Viable Solution? Data privacy Paper: Descher et al., Retaining Data Control in Infrastructure Clouds, ARES (the International Dependability Conference), 2009.

35 Performance Difficult to track in a virtualized environment I/O can be an issue Tradeoffs between CPU power and throughput New networking solutions Several studies of cloud performance One paper: Walker, Benchmarking Amazon EC2 for high-performance scientific computing Low bandwidth from existing providers: 2-5 MB/sec, 17/21 MB/sec, 30MB/sec Generally speaking, the existing cloud providers do not offer a very high-end computer

36 Price Price for what? Experimenting with business models Estimating the cost is hard Price of Base Services for AWS: Computation / EC2 On-demand: starting at $0.1 per hour Reserved: starting at $325 per year for $0.03 per hour Data / S3 Storage: $0.15 per GB/month, Transfer: $0.17 per GB AWS import/export for bulk Hosting Scientific datasets for free Free on AWS for frequently used datasets

37 Service Levels Service levels Computation: immediate, advance reservations, best-effort Data: durability, high/low availability, access performance Cross-cutting concern: security and privacy Different price points for different availability

38 Parting Thoughts IaaS cloud computing is science-driven Scientific applications are successfully using the existing infrastructure for production runs Many more could be using it, but challenges exist Project for the next few years: solve them!