Software Defined Analytics with File and Object Access Plus Geographically Distributed Data
- 1. #ibmedge© 2016 IBM Corporation Software Defined Analytics with File and Object Access Plus Geographically Distributed Data Sandeep Patil, STSM, Spectrum Scale Trishali Nayar, AFM Development, Spectrum Scale Smita Raut, Object Development, Spectrum Scale 22 Sep 2016 Acknowledgement: Bill Owen, Dean Hilderbrand, Sanjay Gandhi, Brian Nelson, Tomonori Kubota, Gyoh Ohsawa
- 2. #ibmedge Agenda • Introduction to Spectrum Scale Active File Manager (AFM) • AFM Use Cases • Spectrum Scale Protocol • Unified File & Object Access (UFO) Feature Details • AFM + Object : Unique Wan Caching for Object Store • Deep Dive on Single Site & Multi-site Caching • Configuration Commands with Demo • Q & A 1
- 3. © 2016 IBM Corporation #ibmedge Spectrum Scale Active File Management (AFM)
- 4. #ibmedge Spectrum Scale –The Complete Data Management Solution 3
- 5. #ibmedge AFM Overview • Active file management (AFM) uses a home-and-cache model in which a single home provides the primary storage of data, and exported data is cached in a local GPFS™ file system • AFM is primarily suited for remote caching • Users access files from the cache system • For read requests, when the file is not yet cached, AFM retrieves the file from the home site • For write requests, writes are allowed on the cache system and can be pushed back to the home system, depending on the cache types 4
- 6. #ibmedge AFM Caching Overview 5 Spectrum Scale Storage Array Storage node Storage node Home Cluster Spectrum Scale Storage Array Storage node Storage node Cache Cluster Nodes are made NFS servers Few nodes are made gateway nodes Cache filesets are associated to NFS export at home.
- 7. #ibmedge Global Sharing with Spectrum Scale AFM • Expands the GPFS global namespace across geographical distances – Caches local ‘copies’ of data distributed to one or more GPFS clusters – Low latency ‘local’ read and write performance – Automated namespace management – As data is written or modified at one location, all other locations see that same data • Efficient data transfers over wide area network (WAN) - Works with unreliable, high latency connections • Speeds data access to collaborators and resources around the world 6 GPFS GPFS GPFS
- 8. #ibmedge AFM Caching Basics • Sites – two sides for a cache relationship • A single home cluster – Presents a fileset that can be cached (export with NFS) – Can be non-GPFS cluster/nodes • One or more cache clusters – Associates a local fileset with the home export • AFM Fileset • Independent fileset with per-inode in xattrs • Data is fetched into the fileset on access (or prefetched on command) • Data written to the fileset is copied back to home • Gateway Node (designation) • Maintains an in-memory queue of pending operations • Moves data between the cache and home clusters • Monitors connectivity to home, switches to disconnected mode on outage, triggers recovery on failure 7
- 9. #ibmedge Spectrum Scale AFM Use Cases 8 Global Namespace • Provides common name space across globally distributed cloud • Persistent scalable cache for remote File System Content distribution • Central site is where data is created, maintained • Branch/edge sites can periodically pre-fetch or pull on demand Content Consolidation Disaster Recovery • Replication of data across WAN with consistency points • Failover and Failback support • Branch offices work on local active data • Master repository maintained centrally • Adv functions – backup etc on central site
- 10. © 2016 IBM Corporation #ibmedge Spectrum Scale Protocol
- 11. #ibmedge Enhanced Protocol Support from 4.1.1 release The Challenge: How can I share my storage infrastructure across all of my legacy and new generation applications? The Solution: • The new IBM Spectrum Scale Protocol Node allows access to data stored in a Spectrum Scale filesystem, using additional access methods and protocols. • The Protocol Node functions are clustered and can support transparent failover for NFS and SWIFT protocols as well as SMB protocols. • Multiprotocol data access from other systems using the following protocols • NFS v3 and v4 • SMB 2 and SMB 3.0 mandatory features / CIFS for Windows support. • OpenStack Swift and S3 API support for object storage. 10
- 12. #ibmedge Adding Protocol Support 11 Administrator Command Line Interface Users NFS SMB/CIFS POSIX Open Stack Swift PN1 Protocol Node Flash Disk Tape ExternalTCP/IPorIBNetwork PN2 PNn … NSD1 Network Shared Disks NSD2 NSDn … Physical Storage IBMSpectrumScaleClusterTCP/IPorIBNetwork Mgmt Nodes Authentication Services keystone Open Stack Cinder SpectrumScaleClusterNodes Elastic Storage Server
- 13. #ibmedge IBM Spectrum Scale Benefits 12 Better performance Eliminate hotspots with massively parallel access to files Sequential I/O with ES greater than 400 GB/s Throughput advantage for parallel streaming workloads, e.g. Tech Computing and Analytics More Storage. More Files. Hyper Scale. Simplified Management Easier management with one global namespace instead of managing islands of NAS arrays, e.g. no need to copy data between compute clusters Integrated policy driven automation Fewer storage administrators required Lower Cost Optimizes storage tiers including flash, disk and tape Increased efficiency and more efficient provisioning due to parallelization and striping technology Remove duplicate copies of data, e.g. run analytics on one copy of data without having to set up a separate silo
- 14. #ibmedge IBM Spectrum Scale – Protocol Integration • Software Offering - protocol support is added to GPFS • Can be configured on existing GPFS clusters or new cluster • Support for Intel and Power Systems • RHEL 7/7.1 – Protocol node requirement – Remaining GPFS nodes can have any supported environment/platform • Use of installation”) also limited to RHEL 7/7.1 • Add support for the following protocols • SMB • NFS • Object (HTTP Rest) • Some cluster nodes are designated as “Protocol Nodes” (aka. CES nodes) • Integrated management of the protocol services • Active-Active clustering • High availability through IP fail-over 13
- 15. #ibmedge IBM Spectrum Scale – Protocol Support 14
- 16. #ibmedge Protocol Support Considerations • Adding Protocol Nodes to GPFS Cluster: • All RHEL7/xServers or All RHEL7/pServers • Not NSD Servers • Protocol Export IPs distributed among the protocol nodes – Different policies for balancing and failback • Management: GUI and CLI • Deployment: Easy Automated Deployment • Flexibility: customer choice of nodes/disks/storage options • Scale: limits for capacity/performance based on GPFS; • CES nodes limits based on protocols enabled • 16 nodes, 3000 connections/node and 20K connections/cluster for SMB • 32 nodes for only NFS or only Object or NFS+Object • Security: root access for cluster management but have sudo access support • Roll your own or combine with Lab Services to meet expectations 15
- 17. © 2016 IBM Corporation #ibmedge Spectrum Scale Object (Part of Spectrum Scale Protocol)
- 18. #ibmedge Spectrum Scale Object Storage • Basic support added in 4.1.1 release & enhanced in 4.2 and 4.2.1 release • Based on Openstack Swift (Juno Release) • REST-based data access • Growing number of clients due to extremely simple protocol • Applications can easily save & access data from anywhere using HTTP • Simple set of atomic operations: – PUT (upload) – POST (update metadata) – GET (download) – DELETE • Amazon S3 Protocol support • High Availability with CES Integration • Simple and Automated Installation Process • Integrated authentication (Keystone) support • Native GPFS Command Line Interface to manage Object service (mmobj command) 17
- 19. #ibmedge Spectrum Scale Object Storage – Additional Features • Unified file and object support with Hadoop connectors • Support for Encryption • Support for Compression • Only Object Store with Tape support for Backup • Object store with integrated transparent cloud tiering Support • Multi Region support • AD/LDAP support for authentication • ILM support for Object • Movement of Object across storage tiers based on access heat • Spectrum Scale Object with IBM DeepFlash becomes object store over all flash array for newer faster workloads. • Spectrum Scale Object with WAN caching support (AFM) 18
- 20. © 2016 IBM Corporation #ibmedge IBM Spectrum Scale: Unified File and Object Access Feature Overview
- 21. #ibmedge Unified File and Object (UFO Support) Spectrum Scale: Redefining Unified Storage • Challenge The world is not converged/file/object/HDFS today! and never will be completely… • Unified Scale-out Content Repository • File or object in. Object or file out. • Integrated big data analytics support • Native protocol support • High-performance that scales • Single Management Plane 20 Spectrum Scale NFS SMBPOSIX SSD Fast Disk Slow Disk Tape Swift/S3HDFS
- 22. #ibmedge What is Unified File and Object Access? • Accessing object using file interfaces (SMB/NFS/POSIX) and accessing file using object interfaces (REST) helps legacy applications designed for file to seamlessly start integrating into the object world. • It allows object data to be accessed using applications designed to process files. It allows file data to be published as objects. • Multi protocol access for file and object in the same namespace (with common User ID management capability) allows supporting and hosting data oceans of different types of data with multiple access options. • Optimizes various use cases and solution architectures resulting in better efficiency as well as cost savings. 21 <Clustered file system> Swift (With Swift on File) NFS/SMB/POSIXObject(http) 2 1 <Container> Data ingested as Objects 3 Data ingested as Files4 Files accessed as Objects
- 23. © 2016 IBM Corporation #ibmedge IBM Spectrum Scale: AFM + Object (Unique Proposition)
- 24. #ibmedge The Need: Thin-Thick storage capacity site deployments for Object Data 23 Applications Applications Applications … Limited storage Limited storage Limited storage Unlimited storageCentral Site Site 3 Site 2 Site 1 Object Data Object Data Object Data Centralized Analytics Centralized Backup • Geo Dispersed multiple sites with limited storage capacity • Independent Applications running on each sites accessing/generating object data. • Centralized Home for consolidated object data – ability to grow storage capacity. • centralized backup for all sites via central location • ability to run analytics for all sites in central location
- 25. #ibmedge Usecase Requirements • There is an object store site that is closer to the end application but has a limited storage capacity. • To cater to large storage capacity requirement there is another object store setup at a geographically remote site which has unlimited or expandable storage capacity, that acts as a central archive. • The relationship between these two object stores need to be setup in such a way that allows applications to access all object data from the site closer to them for faster access, even though it has limited storage capacity. • The central site should have ability to do in place analytics of data. • The central site should have ability to do backup of the data. • If cache goes down the application should be able to failover to the central site. 24
- 26. #ibmedge The Solution: Unique WAN caching for Object Store - available only with Spectrum Scale 25 … Unlimited storage Central Site Centralized Analytics Centralized Backup Applications Limited storage Site 1 Object Data Spectrum Scale Cluster with Protocol Nodes (Object Enabled) Spectrum Scale Cluster with Protocol Nodes (Object Enabled) Spectrum Scale AFM (IW) Relationship with cache eviction enabled on Site 1 Object Data can be accessed as Files using Unified file and Object Feature and used for analytics Data can be centrally backed to TapeSpectrum Scale Feature Requirements Addressed AFM with Spectrum Scale Object - Allows objects store to have thin cache with eviction enabled and thick home. AFM in IW Modes Allows for fail-back and fail-over from cache site to Home useful during disaster. Unified File and Object Access with HDFS connector Allows centralized and in-place analytics of data at Home site Tape Integration Centralized backup
- 27. #ibmedge Thin Object Store Cache – Thick Object Store Archive 26 Spectrum Scale Home#1 Spectrum Scale Cache#1Service 1 Serives XXX Site #1 Fileset Object access Object Ingest Fileset 11TB/d ay AFM Independent-Writer Swift API Swift API Failover/Failback Existing Services Cache in Region 1 Archive in Region 2 Replicate XXTB of data everyday • Cache Site in Region 1 with limited storage and Home site in Region with max storage per data center • Object data to be archived from cache site in Region 1 to home site in Region 2 using AFM –IW • On cache failure, application will fail over home site for object access. Application will fail-back when cache comes up. • Limited storage on cache site addressed by using Eviction along with AFM • Key Features used in Solution: Spectrum Scale Object , AFM (IW) with Eviction • Available and documented in 4.2.1
- 28. #ibmedge Spectrum Scale Cluster for Region 1 Home Cluster for Region 1 Service s Service s Region #1 Spectrum Scale Cluster for Region 1 Service s Service s Region #2 SwiftAPI Objects Objects Existing Services Cache Home in Region 3 Home Cluster for Region 2 Swift API Swift API Failover/Failback Swift API Swift APIFailover/Failback One can include multiple sites where each site has its own home cluster at the central region and replicate the setup shown in previous slide for single site. Multiple site Deployment
- 29. #ibmedge Configuration Steps • Details Configuration Step Available in 4.2.1 in Knowledge Center Using AFM with Spectrum Scale Object • http://www.ibm.com/support/knowledgecenter/STXKQY_4.2.1/com.ibm.s pectrum.scale.v4r21.doc/bl1ins_usingafmwithobject.htm 28
- 30. #ibmedge Conclusion • Spectrum Scale provides rich set of features like • AFM • Protocols with POSIX, SMB,NFS and Object • Unified File and Object Access • In Place analytics using build-in Hadoop connectors • Integrating AFM with spectrum scale object delivers unique solution required for many multi-site deployments wherein: • One can have thin cache object store with auto eviction facility closer to the applications or users • Centralized thick home object store which can act as failback object store for the thin cache sites. • Ability to do in-place analytics of all the data on the home site • Ability to do a central backup at the home site. 29
- 31. #ibmedge Spectrum Scale User Group • The Spectrum Scale User Group is free to join and open to all using, interested in using or integrating Spectrum Scale. • Join the User Group activities to meet your peers and get access to experts from partners and IBM. • Driven and owned by Customers • Next meetings: - APAC: October 14, Melbourne - Global at SC16 : November 13 1pm to 5pm, Salt Lake City • Web page: http://www.spectrumscale.org/ • Presentations: http://www.spectrumscale.org/presentations/ • Mailing list: http://www.spectrumscale.org/join/ • Contact: http://www.spectrumscale.org/committee/ • Meet Bob Oesterlin (US Co-Principal) at Edge2016: [email protected]
- 32. #ibmedge Session : How to apply Flash benefits to big data analytics and unstructured data NDA & Customers ONLY • Who: IBM Elastic Storage Server Offering Management • Alex Chen • When: Thursday, September 22, 2016 • 1:15pm to 2:15pm • Where: Grand Garden Arena, Lower Level, MGM, Studio 10 • Contact(if any questions) • • [email protected], [email protected] 31
- 33. #ibmedge Spectrum Scale Trial VM • Download the IBM Spectrum Scale Trial VM from : • http://www-03.ibm.com/systems/storage/spectrum/scale/trial.html 32
- 34. #ibmedge References Spectrum Scale 4.2.1 Knowledge Center: Using AFM with object http://www.ibm.com/support/knowledgecenter/STXKQY_4.2.1/com.ibm.spectrum.scale.v4r21.doc/bl1ins_usingafm withobject.htm Spectrum Scale Object Store – Unified File and Object http://www.slideshare.net/SandeepPatil154/spectrum-scaleexternalunifiedfile-object From Archive to Insight: Debunking Myths of Analytics on Object Stores – Dean Hildebrand, Bill Owen, Simon Lorenz, Luis Pabon, Rui Zhang. Vancouver Summit, Spring 2015. https://www.youtube.com/watch?v=brhEUptD3JQ Deploying Swift on a File System – Bill Owen, Thiago Da Silva. BrownBag at OpenStack Paris, Fall 2014 https://www.youtube.com/watch?v=vPn2uZF4yWo Breaking the Mold with OpenStack Swift and GlusterFS – Jon Dickinson, Luis Pabo. Atlanta Summit, Spring 2014 https://www.youtube.com/watch?v=pSWdzjA8WuA SNIA SDC 2015 http://www.snia.org/sites/default/files/SDC15_presentations/security/DeanHildebrand_Sasi__OpenStack%20Swift OnFile.pdf
- 35. #ibmedge Notices and Disclaimers 34 Copyright © 2016 by International Business Machines Corporation (IBM). No part of this document may be reproduced or transmitted in any form without written permission from IBM. U.S. Government Users Restricted Rights - Use, duplication or disclosure restricted by GSA ADP Schedule Contract with IBM. Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors. IBM shall have no responsibility to update this information. THIS DOCUMENT IS DISTRIBUTED "AS IS" WITHOUT ANY WARRANTY, EITHER EXPRESS OR IMPLIED. IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION, INCLUDING BUT NOT LIMITED TO, LOSS OF DATA, BUSINESS INTERRUPTION, LOSS OF PROFIT OR LOSS OF OPPORTUNITY. IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided. IBM products are manufactured from new parts or new and used parts. In some cases, a product may not be new and may have been previously installed. Regardless, our warranty terms apply.” Any statements regarding IBM's future direction, intent or product plans are subject to change or withdrawal without notice. Performance data contained herein was generally obtained in a controlled, isolated environments. Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved. Actual performance, cost, savings or other results in other operating environments may vary. References in this document to IBM products, programs, or services does not imply that IBM intends to make such products, programs or services available in all countries in which IBM operates or does business. Workshops, sessions and associated materials may have been prepared by independent session speakers, and do not necessarily reflect the views of IBM. All materials and discussions are provided for informational purposes only, and are neither intended to, nor shall constitute legal or other guidance or advice to any individual participant or their specific situation. It is the customer’s responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customer’s business and any actions the customer may need to take to comply with such laws. IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
- 36. #ibmedge Notices and Disclaimers Con’t. 35 Information concerning non-IBM products was obtained from the suppliers of those products, their published announcements or other publicly available sources. IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance, compatibility or any other claims related to non-IBM products. Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products. IBM does not warrant the quality of any third-party products, or the ability of any such third-party products to interoperate with IBM’s products. IBM EXPRESSLY DISCLAIMS ALL WARRANTIES, EXPRESSED OR IMPLIED, INCLUDING BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. The provision of the information contained h erein is not intended to, and does not, grant any right or license under any IBM patents, copyrights, trademarks or other intellectual property right. IBM, the IBM logo, ibm.com, Aspera®, Bluemix, Blueworks Live, CICS, Clearcase, Cognos®, DOORS®, Emptoris®, Enterprise Document Management System™, FASP®, FileNet®, Global Business Services ®, Global Technology Services ®, IBM ExperienceOne™, IBM SmartCloud®, IBM Social Business®, Information on Demand, ILOG, Maximo®, MQIntegrator®, MQSeries®, Netcool®, OMEGAMON, OpenPower, PureAnalytics™, PureApplication®, pureCluster™, PureCoverage®, PureData®, PureExperience®, PureFlex®, pureQuery®, pureScale®, PureSystems®, QRadar®, Rational®, Rhapsody®, Smarter Commerce®, SoDA, SPSS, Sterling Commerce®, StoredIQ, Tealeaf®, Tivoli®, Trusteer®, Unica®, urban{code}®, Watson, WebSphere®, Worklight®, X-Force® and System z® Z/OS, are trademarks of International Business Machines Corporation, registered in many jurisdictions worldwide. Other product and service names might be trademarks of IBM or other companies. A current list of IBM trademarks is available on the Web at "Copyright and trademark information" at: www.ibm.com/legal/copytrade.shtml.
- 37. #ibmedge IBM Spectrum Scale Summary 36 • Avoid vendor lock-in with true Software Avoid vendor lock-in with true Software Defined Storage and Open Standards • Seamless performance & capacity scaling • Automate data management at scale • Enable global collaboration Data management at scale OpenStack and Spectrum Scale helps clients manage data at scale Business: I need virtually unlimited storage Operations: I need a flexible infrastructure that supports both object and file based storage Operations: I need to minimize the time it takes to perform common storage management tasks A single data plane that supports Cinder, Glance, Swift, Manila as well as NFS, SMB, et. al. A fully automated policy based data placement and migration tool An open & scalable cloud platform Sharing with a variety of WAN caching modes Results • Converge File and Object based storage under one roof • Employ enterprise features to protect data, e.g. Snapshots, Backup, and Disaster Recovery • Support native file, block and object sharing to data Spectrum Scale NFS SMBPOSIX SSD Fast Disk Slow Disk Tape Swift HDFS Cinder Glance Manila 36 Collaboration: I need to share data between people, departments and sites with low latency. Data management at scale
- 38. © 2016 IBM Corporation #ibmedge Thank You